abstract
stringlengths 13
4.33k
| field
sequence | task
sequence | method
sequence | dataset
sequence | metric
sequence | title
stringlengths 10
194
|
|---|---|---|---|---|---|---|
Adversarial examples are commonly viewed as a threat to ConvNets. Here we present an opposite perspective: adversarial examples can be used to improve image recognition models if harnessed in the right manner. We propose AdvProp, an enhanced adversarial training scheme which treats adversarial examples as additional examples, to prevent overfitting. Key to our method is the usage of a separate auxiliary batch norm for adversarial examples, as they have different underlying distributions to normal examples. We show that AdvProp improves a wide range of models on various image recognition tasks and performs better when the models are bigger. For instance, by applying AdvProp to the latest EfficientNet-B7 [28] on ImageNet, we achieve significant improvements on ImageNet (+0.7%), ImageNet-C (+6.5%), ImageNet-A (+7.0%), Stylized-ImageNet (+4.8%). With an enhanced EfficientNet-B8, our method achieves the state-of-the-art 85.5% ImageNet top-1 accuracy without extra data. This result even surpasses the best model in [20] which is trained with 3.5B Instagram images (~3000X more than ImageNet) and ~9.4X more parameters. Models are available at https://github.com/tensorflow/tpu/tree/master/models/official/efficientnet. | [
"Image Data Augmentation",
"Regularization",
"Adversarial Training",
"Stochastic Optimization",
"Recurrent Neural Networks",
"Activation Functions",
"Normalization",
"Convolutions",
"Feedforward Networks",
"Pooling Operations",
"Image Model Blocks",
"Image Models",
"Skip Connection Blocks"
] | [
"Image Classification"
] | [
"Depthwise Convolution",
"Weight Decay",
"Average Pooling",
"EfficientNet",
"RMSProp",
"Cutout",
"Long Short-Term Memory",
"Tanh Activation",
"1x1 Convolution",
"Auxiliary Batch Normalization",
"AutoAugment",
"Convolution",
"ReLU",
"Dense Connections",
"Swish",
"Batch Normalization",
"AdvProp",
"ColorJitter",
"Squeeze-and-Excitation Block",
"Pointwise Convolution",
"Sigmoid Activation",
"Color Jitter",
"Inverted Residual Block",
"LSTM",
"Dropout",
"Depthwise Separable Convolution",
"Rectified Linear Units"
] | [
"ImageNet"
] | [
"Number of params",
"Top 5 Accuracy",
"Top 1 Accuracy"
] | Adversarial Examples Improve Image Recognition |
We propose HOI Transformer to tackle human object interaction (HOI) detection in an end-to-end manner. Current approaches either decouple HOI task into separated stages of object detection and interaction classification or introduce surrogate interaction problem. In contrast, our method, named HOI Transformer, streamlines the HOI pipeline by eliminating the need for many hand-designed components. HOI Transformer reasons about the relations of objects and humans from global image context and directly predicts HOI instances in parallel. A quintuple matching loss is introduced to force HOI predictions in a unified way. Our method is conceptually much simpler and demonstrates improved accuracy. Without bells and whistles, HOI Transformer achieves $26.61\% $ $ AP $ on HICO-DET and $52.9\%$ $AP_{role}$ on V-COCO, surpassing previous methods with the advantage of being much simpler. We hope our approach will serve as a simple and effective alternative for HOI tasks. Code is available at https://github.com/bbepoch/HoiTransformer . | [
"Transformers"
] | [
"Human-Object Interaction Detection",
"Object Detection"
] | [
"Transformer"
] | [
"HICO-DET"
] | [
"MAP"
] | End-to-End Human Object Interaction Detection with HOI Transformer |
Graph convolutional neural networks (GCNs) embed nodes in a graph into Euclidean space, which has been shown to incur a large distortion when embedding real-world graphs with scale-free or hierarchical structure. Hyperbolic geometry offers an exciting alternative, as it enables embeddings with much smaller distortion. However, extending GCNs to hyperbolic geometry presents several unique challenges because it is not clear how to define neural network operations, such as feature transformation and aggregation, in hyperbolic space. Furthermore, since input features are often Euclidean, it is unclear how to transform the features into hyperbolic embeddings with the right amount of curvature. Here we propose Hyperbolic Graph Convolutional Neural Network (HGCN), the first inductive hyperbolic GCN that leverages both the expressiveness of GCNs and hyperbolic geometry to learn inductive node representations for hierarchical and scale-free graphs. We derive GCN operations in the hyperboloid model of hyperbolic space and map Euclidean input features to embeddings in hyperbolic spaces with different trainable curvature at each layer. Experiments demonstrate that HGCN learns embeddings that preserve hierarchical structure, and leads to improved performance when compared to Euclidean analogs, even with very low dimensional embeddings: compared to state-of-the-art GCNs, HGCN achieves an error reduction of up to 63.1% in ROC AUC for link prediction and of up to 47.5% in F1 score for node classification, also improving state-of-the art on the Pubmed dataset. | [
"Graph Models"
] | [
"Hierarchical structure",
"Link Prediction",
"Node Classification"
] | [
"Graph Convolutional Network",
"GCN"
] | [
"Cora",
"Pubmed",
"PPI"
] | [
"Accuracy"
] | Hyperbolic Graph Convolutional Neural Networks |
We present Momentum Contrast (MoCo) for unsupervised visual representation learning. From a perspective on contrastive learning as dictionary look-up, we build a dynamic dictionary with a queue and a moving-averaged encoder. This enables building a large and consistent dictionary on-the-fly that facilitates contrastive unsupervised learning. MoCo provides competitive results under the common linear protocol on ImageNet classification. More importantly, the representations learned by MoCo transfer well to downstream tasks. MoCo can outperform its supervised pre-training counterpart in 7 detection/segmentation tasks on PASCAL VOC, COCO, and other datasets, sometimes surpassing it by large margins. This suggests that the gap between unsupervised and supervised representation learning has been largely closed in many vision tasks. | [
"Convolutional Neural Networks",
"Feature Extractors",
"Normalization",
"Self-Supervised Learning",
"Activation Functions",
"Convolutions",
"Pooling Operations",
"Object Detection Models",
"Region Proposal",
"Stochastic Optimization",
"Loss Functions",
"Skip Connection Blocks",
"Image Data Augmentation",
"Initialization",
"Output Functions",
"Learning Rate Schedules",
"RoI Feature Extractors",
"Instance Segmentation Models",
"Skip Connections"
] | [
"Representation Learning",
"Self-Supervised Image Classification"
] | [
"InfoNCE",
"Faster R-CNN",
"Average Pooling",
"1x1 Convolution",
"RoIAlign",
"Region Proposal Network",
"ResNet",
"MoCo",
"Random Horizontal Flip",
"RoIPool",
"Convolution",
"ReLU",
"Residual Connection",
"FPN",
"Random Grayscale",
"RPN",
"Momentum Contrast",
"Random Resized Crop",
"Batch Normalization",
"Residual Network",
"ColorJitter",
"Kaiming Initialization",
"Exponential Decay",
"Color Jitter",
"SGD with Momentum",
"Softmax",
"Feature Pyramid Network",
"Bottleneck Residual Block",
"Mask R-CNN",
"Residual Block",
"Global Average Pooling",
"Rectified Linear Units",
"Max Pooling"
] | [
"ImageNet"
] | [
"Top 1 Accuracy (kNN, k=20)",
"Number of Params",
"Top 1 Accuracy"
] | Momentum Contrast for Unsupervised Visual Representation Learning |
We investigate the recently proposed Time-domain Audio Sep-aration Network (TasNet) in the task of real-time single-channel speech dereverberation. Unlike systems that take time-frequency representation of the audio as input, TasNet learns anadaptive front-end in replacement of the time-frequency rep-resentation by a time-domain convolutional non-negative au-toencoder. We show that by formulating the dereverberationproblem as a denoising problem where the direct path is sepa-rated from the reverberations, a TasNet denoising autoencodercan outperform a deep LSTM baseline on log-power magnitudespectrogram input in both causal and non-causal settings. Wefurther show that adjusting the stride size in the convolutionalautoencoder helps both the dereverberation and separation per-formance. | [
"Recurrent Neural Networks",
"Activation Functions"
] | [
"Denoising",
"Speech Dereverberation",
"Speech Separation"
] | [
"Tanh Activation",
"Long Short-Term Memory",
"LSTM",
"Sigmoid Activation"
] | [
"wsj0-2mix"
] | [
"SI-SDRi"
] | Real-time Single-channel Dereverberation and Separation with Time-domainAudio Separation Network |
Vehicle 3D extents and trajectories are critical cues for predicting the future location of vehicles and planning future agent ego-motion based on those predictions. In this paper, we propose a novel online framework for 3D vehicle detection and tracking from monocular videos. The framework can not only associate detections of vehicles in motion over time, but also estimate their complete 3D bounding box information from a sequence of 2D images captured on a moving platform. Our method leverages 3D box depth-ordering matching for robust instance association and utilizes 3D trajectory prediction for re-identification of occluded vehicles. We also design a motion learning module based on an LSTM for more accurate long-term motion extrapolation. Our experiments on simulation, KITTI, and Argoverse datasets show that our 3D tracking pipeline offers robust data association and tracking. On Argoverse, our image-based method is significantly better for tracking 3D vehicles within 30 meters than the LiDAR-centric baseline methods. | [
"Recurrent Neural Networks",
"Activation Functions"
] | [
"3D Object Detection",
"3D Pose Estimation",
"Autonomous Vehicles",
"Multiple Object Tracking",
"Object Tracking",
"Online Multi-Object Tracking",
"Pose Estimation",
"Trajectory Prediction"
] | [
"Tanh Activation",
"Long Short-Term Memory",
"LSTM",
"Sigmoid Activation"
] | [
"KITTI Tracking test"
] | [
"MOTA"
] | Joint Monocular 3D Vehicle Detection and Tracking |
Lung cancer classification in screening computed tomography (CT) scans is one of the most crucial tasks for early detection of this disease. Many lives can be saved if we are able to accurately classify malignant/ cancerous lung nodules. Consequently, several deep learning based models have been proposed recently to classify lung nodules as malignant or benign. Nevertheless, the large variation in the size and heterogeneous appearance of the nodules makes this task an extremely challenging one. We propose a new Progressive Growing Channel Attentive Non-Local (ProCAN) network for lung nodule classification. The proposed method addresses this challenge from three different aspects. First, we enrich the Non-Local network by adding channel-wise attention capability to it. Second, we apply Curriculum Learning principles, whereby we first train our model on easy examples before hard/ difficult ones. Third, as the classification task gets harder during the Curriculum learning, our model is progressively grown to increase its capability of handling the task at hand. We examined our proposed method on two different public datasets and compared its performance with state-of-the-art methods in the literature. The results show that the ProCAN model outperforms state-of-the-art methods and achieves an AUC of 98.05% and accuracy of 95.28% on the LIDC-IDRI dataset. Moreover, we conducted extensive ablation studies to analyze the contribution and effects of each new component of our proposed method. | [
"Attention Mechanisms"
] | [
"Computed Tomography (CT)",
"Curriculum Learning",
"Lung Nodule Classification"
] | [
"Progressive Growing Channel Attentive Non-Local Network",
"ProCAN"
] | [
"LIDC-IDRI"
] | [
"AUC",
"Accuracy"
] | ProCAN: Progressive Growing Channel Attentive Non-Local Network for Lung Nodule Classification |
We present a Parallel Iterative Edit (PIE) model for the problem of local sequence transduction arising in tasks like Grammatical error correction (GEC). Recent approaches are based on the popular encoder-decoder (ED) model for sequence to sequence learning. The ED model auto-regressively captures full dependency among output tokens but is slow due to sequential decoding. The PIE model does parallel decoding, giving up the advantage of modelling full dependency in the output, yet it achieves accuracy competitive with the ED model for four reasons: 1.~predicting edits instead of tokens, 2.~labeling sequences instead of generating sequences, 3.~iteratively refining predictions to capture dependencies, and 4.~factorizing logits over edits and their token argument to harness pre-trained language models like BERT. Experiments on tasks spanning GEC, OCR correction and spell correction demonstrate that the PIE model is an accurate and significantly faster alternative for local sequence transduction. | [
"Output Functions",
"Regularization",
"Stochastic Optimization",
"Attention Modules",
"Learning Rate Schedules",
"Activation Functions",
"Normalization",
"Subword Segmentation",
"Language Models",
"Feedforward Networks",
"Attention Mechanisms",
"Skip Connections"
] | [
"Grammatical Error Correction",
"Optical Character Recognition"
] | [
"Weight Decay",
"Layer Normalization",
"WordPiece",
"Softmax",
"Adam",
"Multi-Head Attention",
"Attention Dropout",
"Linear Warmup With Linear Decay",
"Residual Connection",
"Scaled Dot-Product Attention",
"Dropout",
"BERT",
"GELU",
"Dense Connections",
"Gaussian Linear Error Units"
] | [
"CoNLL-2014 Shared Task"
] | [
"F0.5"
] | Parallel Iterative Edit Models for Local Sequence Transduction |
We consider the problem of detecting out-of-distribution images in neural networks. We propose ODIN, a simple and effective method that does not require any change to a pre-trained neural network. Our method is based on the observation that using temperature scaling and adding small perturbations to the input can separate the softmax score distributions between in- and out-of-distribution images, allowing for more effective detection. We show in a series of experiments that ODIN is compatible with diverse network architectures and datasets. It consistently outperforms the baseline approach by a large margin, establishing a new state-of-the-art performance on this task. For example, ODIN reduces the false positive rate from the baseline 34.7% to 4.3% on the DenseNet (applied to CIFAR-10) when the true positive rate is 95%. | [
"Initialization",
"Regularization",
"Convolutional Neural Networks",
"Output Functions",
"Activation Functions",
"Normalization",
"Convolutions",
"Feedforward Networks",
"Pooling Operations",
"Skip Connections",
"Image Model Blocks"
] | [
"Out-of-Distribution Detection"
] | [
"Dense Block",
"Average Pooling",
"Softmax",
"Concatenated Skip Connection",
"Convolution",
"Batch Normalization",
"1x1 Convolution",
"ReLU",
"Dropout",
"DenseNet",
"Kaiming Initialization",
"Dense Connections",
"Global Average Pooling",
"Rectified Linear Units",
"Max Pooling"
] | [
"MS-1M vs. IJB-C",
"ImageNet dogs vs ImageNet non-dogs"
] | [
"AUROC"
] | Enhancing The Reliability of Out-of-distribution Image Detection in Neural Networks |
Offline training for object tracking has recently shown great potentials in balancing tracking accuracy and speed. However, it is still difficult to adapt an offline trained model to a target tracked online. This work presents a Residual Attentional Siamese Network (RASNet) for high performance object tracking. The RASNet model reformulates the correlation filter within a Siamese tracking framework, and introduces different kinds of the attention mechanisms to adapt the model without updating the model online. In particular, by exploiting the offline trained general attention, the target adapted residual attention, and the channel favored feature attention, the RASNet not only mitigates the over-fitting problem in deep network training, but also enhances its discriminative capacity and adaptability due to the separation of representation learning and discriminator learning. The proposed deep architecture is trained from end to end and takes full advantage of the rich spatial temporal information to achieve robust visual tracking. Experimental results on two latest benchmarks, OTB-2015 and VOT2017, show that the RASNet tracker has the state-of-the-art tracking accuracy while runs at more than 80 frames per second. | [
"Twin Networks"
] | [
"Object Tracking",
"Representation Learning",
"Visual Tracking"
] | [
"Siamese Network"
] | [
"OTB-2013"
] | [
"AUC"
] | Learning Attentions: Residual Attentional Siamese Network for High Performance Online Visual Tracking |
Recently, the character-word lattice structure has been proved to be effective for Chinese named entity recognition (NER) by incorporating the word information. However, since the lattice structure is complex and dynamic, most existing lattice-based models are hard to fully utilize the parallel computation of GPUs and usually have a low inference-speed. In this paper, we propose FLAT: Flat-LAttice Transformer for Chinese NER, which converts the lattice structure into a flat structure consisting of spans. Each span corresponds to a character or latent word and its position in the original lattice. With the power of Transformer and well-designed position encoding, FLAT can fully leverage the lattice information and has an excellent parallelization ability. Experiments on four datasets show FLAT outperforms other lexicon-based models in performance and efficiency. | [
"Regularization",
"Output Functions",
"Stochastic Optimization",
"Attention Modules",
"Activation Functions",
"Subword Segmentation",
"Normalization",
"Feedforward Networks",
"Transformers",
"Attention Mechanisms",
"Skip Connections"
] | [
"Chinese Named Entity Recognition",
"Named Entity Recognition"
] | [
"Layer Normalization",
"Byte Pair Encoding",
"BPE",
"Softmax",
"Adam",
"Transformer",
"Multi-Head Attention",
"Rectified Linear Units",
"ReLU",
"Residual Connection",
"Label Smoothing",
"Dropout",
"Scaled Dot-Product Attention",
"Dense Connections"
] | [
"Resume NER",
"MSRA",
"OntoNotes 4",
"Weibo NER"
] | [
"F1"
] | FLAT: Chinese NER Using Flat-Lattice Transformer |
Popular graph neural networks implement convolution operations on graphs based on polynomial spectral filters. In this paper, we propose a novel graph convolutional layer inspired by the auto-regressive moving average (ARMA) filter that, compared to polynomial ones, provides a more flexible frequency response, is more robust to noise, and better captures the global graph structure. We propose a graph neural network implementation of the ARMA filter with a recursive and distributed formulation, obtaining a convolutional layer that is efficient to train, localized in the node space, and can be transferred to new graphs at test time. We perform a spectral analysis to study the filtering effect of the proposed ARMA layer and report experiments on four downstream tasks: semi-supervised node classification, graph signal classification, graph classification, and graph regression. Results show that the proposed ARMA layer brings significant improvements over graph neural networks based on polynomial filters. | [
"Convolutions"
] | [
"Graph Classification",
"Graph Regression",
"Node Classification",
"Regression",
"Skeleton Based Action Recognition"
] | [
"Convolution"
] | [
"SBU",
"Lipophilicity "
] | [
"RMSE",
"Accuracy"
] | Graph Neural Networks with convolutional ARMA filters |
GANs excel at learning high dimensional distributions, but they can update
generator parameters in directions that do not correspond to the steepest
descent direction of the objective. Prominent examples of problematic update
directions include those used in both Goodfellow's original GAN and the
WGAN-GP. To formally describe an optimal update direction, we introduce a
theoretical framework which allows the derivation of requirements on both the
divergence and corresponding method for determining an update direction, with
these requirements guaranteeing unbiased mini-batch updates in the direction of
steepest descent. We propose a novel divergence which approximates the
Wasserstein distance while regularizing the critic's first order information.
Together with an accompanying update direction, this divergence fulfills the
requirements for unbiased steepest descent updates. We verify our method, the
First Order GAN, with image generation on CelebA, LSUN and CIFAR-10 and set a
new state of the art on the One Billion Word language generation task. Code to
reproduce experiments is available. | [
"Generative Models",
"Convolutions"
] | [
"Image Generation",
"Text Generation"
] | [
"Generative Adversarial Network",
"GAN",
"Convolution"
] | [
"LSUN Bedroom 64 x 64",
"CIFAR-10"
] | [
"FID"
] | First Order Generative Adversarial Networks |
We observed that recent state-of-the-art results on single image human pose
estimation were achieved by multi-stage Convolution Neural Networks (CNN).
Notwithstanding the superior performance on static images, the application of
these models on videos is not only computationally intensive, it also suffers
from performance degeneration and flicking. Such suboptimal results are mainly
attributed to the inability of imposing sequential geometric consistency,
handling severe image quality degradation (e.g. motion blur and occlusion) as
well as the inability of capturing the temporal correlation among video frames.
In this paper, we proposed a novel recurrent network to tackle these problems.
We showed that if we were to impose the weight sharing scheme to the
multi-stage CNN, it could be re-written as a Recurrent Neural Network (RNN).
This property decouples the relationship among multiple network stages and
results in significantly faster speed in invoking the network for videos. It
also enables the adoption of Long Short-Term Memory (LSTM) units between video
frames. We found such memory augmented RNN is very effective in imposing
geometric consistency among frames. It also well handles input quality
degradation in videos while successfully stabilizes the sequential outputs. The
experiments showed that our approach significantly outperformed current
state-of-the-art methods on two large-scale video pose estimation benchmarks.
We also explored the memory cells inside the LSTM and provided insights on why
such mechanism would benefit the prediction for video-based pose estimations. | [
"Convolutions",
"Activation Functions",
"Recurrent Neural Networks"
] | [
"Pose Estimation"
] | [
"Long Short-Term Memory",
"Convolution",
"Tanh Activation",
"LSTM",
"Sigmoid Activation"
] | [
"UPenn Action",
"J-HMDB"
] | [
"Mean [email protected]"
] | LSTM Pose Machines |
Recent work in network quantization produced state-of-the-art results using mixed precision quantization. An imperative requirement for many efficient edge device hardware implementations is that their quantizers are uniform and with power-of-two thresholds. In this work, we introduce the Hardware Friendly Mixed Precision Quantization Block (HMQ) in order to meet this requirement. The HMQ is a mixed precision quantization block that repurposes the Gumbel-Softmax estimator into a smooth estimator of a pair of quantization parameters, namely, bit-width and threshold. HMQs use this to search over a finite space of quantization schemes. Empirically, we apply HMQs to quantize classification models trained on CIFAR10 and ImageNet. For ImageNet, we quantize four different architectures and show that, in spite of the added restrictions to our quantization scheme, we achieve competitive and, in some cases, state-of-the-art results. | [
"Distributions",
"Stochastic Optimization"
] | [
"Quantization"
] | [
"Gumbel Softmax",
"RAdam"
] | [
"ImageNet"
] | [
"Accuracy (%)"
] | HMQ: Hardware Friendly Mixed Precision Quantization Block for CNNs |
Suffering from the extreme training data imbalance between seen and unseen
classes, most of existing state-of-the-art approaches fail to achieve
satisfactory results for the challenging generalized zero-shot learning task.
To circumvent the need for labeled examples of unseen classes, we propose a
novel generative adversarial network (GAN) that synthesizes CNN features
conditioned on class-level semantic information, offering a shortcut directly
from a semantic descriptor of a class to a class-conditional feature
distribution. Our proposed approach, pairing a Wasserstein GAN with a
classification loss, is able to generate sufficiently discriminative CNN
features to train softmax classifiers or any multimodal embedding method. Our
experimental results demonstrate a significant boost in accuracy over the state
of the art on five challenging datasets -- CUB, FLO, SUN, AWA and ImageNet --
in both the zero-shot learning and generalized zero-shot learning settings. | [
"Generative Models",
"Convolutions",
"Output Functions"
] | [
"Generalized Zero-Shot Learning",
"Zero-Shot Learning"
] | [
"Softmax",
"Generative Adversarial Network",
"GAN",
"Convolution"
] | [
"SUN Attribute",
"CUB-200-2011"
] | [
"average top-1 classification accuracy",
"Harmonic mean"
] | Feature Generating Networks for Zero-Shot Learning |
This paper presents an approach to perform human activity recognition in videos through the employment of a deep recurrent network, taking as inputs appearance and optical flow information. Our method proposes a novel architecture named BubbleNET, which is based on a recurrent layer dispersed into several modules (referred to as bubbles) along with an attention mechanism based on squeeze-and-excitation strategy, responsible to modulate each bubble contribution. Thereby, we intend to gather information from fundamentally correlated segments of the input data, creating a signature of components that characterize each activity. Our experiments, conducted on widely employed activity recognition datasets, support the existence of these signatures, evidenced by maps of bubble activations for every class of the datasets. To compare the approach to literature methods, mean accuracy is taken into account, for which BubbleNET obtained 97.62%, 91.70% and 82.60% on UCF-101, YUP++ and HMDB-51 datasets, respectively, being placed among state-of-the-art methods. | [
"Activation Functions",
"Convolutions",
"Feedforward Networks",
"Pooling Operations",
"Image Model Blocks"
] | [
"Action Recognition",
"Activity Recognition",
"Activity Recognition In Videos",
"Optical Flow Estimation"
] | [
"Average Pooling",
"Convolution",
"ReLU",
"Squeeze-and-Excitation Block",
"Dense Connections",
"Rectified Linear Units",
"Sigmoid Activation"
] | [
"UCF101",
"HMDB-51"
] | [
"Average accuracy of 3 splits",
"3-fold Accuracy"
] | Bubblenet: A Disperse Recurrent Structure To Recognize Activities |
Subword segmentation is widely used to address the open vocabulary problem in machine translation. The dominant approach to subword segmentation is Byte Pair Encoding (BPE), which keeps the most frequent words intact while splitting the rare ones into multiple tokens. While multiple segmentations are possible even with the same vocabulary, BPE splits words into unique sequences; this may prevent a model from better learning the compositionality of words and being robust to segmentation errors. So far, the only way to overcome this BPE imperfection, its deterministic nature, was to create another subword segmentation algorithm (Kudo, 2018). In contrast, we show that BPE itself incorporates the ability to produce multiple segmentations of the same word. We introduce BPE-dropout - simple and effective subword regularization method based on and compatible with conventional BPE. It stochastically corrupts the segmentation procedure of BPE, which leads to producing multiple segmentations within the same fixed BPE framework. Using BPE-dropout during training and the standard BPE during inference improves translation quality up to 3 BLEU compared to BPE and up to 0.9 BLEU compared to the previous subword regularization. | [
"Subword Segmentation"
] | [
"Machine Translation"
] | [
"BPE",
"Byte Pair Encoding"
] | [
"IWSLT2017 Arabic-English",
"IWSLT2015 English-Vietnamese",
"IWSLT2017 English-Arabic",
"IWSLT2017 English-French",
"IWSLT2017 French-English"
] | [
"Cased sacreBLEU",
"BLEU"
] | BPE-Dropout: Simple and Effective Subword Regularization |
Real-noise denoising is a challenging task because the statistics of real-noise do not follow the normal distribution, and they are also spatially and temporally changing. In order to cope with various and complex real-noise, we propose a well-generalized denoising architecture and a transfer learning scheme. Specifically, we adopt an adaptive instance normalization to build a denoiser, which can regularize the feature map and prevent the network from overfitting to the training set. We also introduce a transfer learning scheme that transfers knowledge learned from synthetic-noise data to the real-noise denoiser. From the proposed transfer learning, the synthetic-noise denoiser can learn general features from various synthetic-noise data, and the real-noise denoiser can learn the real-noise characteristics from real data. From the experiments, we find that the proposed denoising method has great generalization ability, such that our network trained with synthetic-noise achieves the best performance for Darmstadt Noise Dataset (DND) among the methods from published papers. We can also see that the proposed transfer learning scheme robustly works for real-noise images through the learning with a very small number of labeled data. | [
"Normalization"
] | [
"Denoising",
"Image Denoising",
"Transfer Learning"
] | [
"Adaptive Instance Normalization"
] | [
"SIDD",
"DND"
] | [
"SSIM (sRGB)",
"PSNR (sRGB)"
] | Transfer Learning from Synthetic to Real-Noise Denoising with Adaptive Instance Normalization |
Skeleton-based Human Activity Recognition has achieved great interest in recent years as skeleton data has demonstrated being robust to illumination changes, body scales, dynamic camera views, and complex background. In particular, Spatial-Temporal Graph Convolutional Networks (ST-GCN) demonstrated to be effective in learning both spatial and temporal dependencies on non-Euclidean data such as skeleton graphs. Nevertheless, an effective encoding of the latent information underlying the 3D skeleton is still an open problem, especially when it comes to extracting effective information from joint motion patterns and their correlations. In this work, we propose a novel Spatial-Temporal Transformer network (ST-TR) which models dependencies between joints using the Transformer self-attention operator. In our ST-TR model, a Spatial Self-Attention module (SSA) is used to understand intra-frame interactions between different body parts, and a Temporal Self-Attention module (TSA) to model inter-frame correlations. The two are combined in a two-stream network, whose performance is evaluated on three large-scale datasets, NTU-RGB+D 60, NTU-RGB+D 120, and Kinetics Skeleton 400, outperforming the state-of-the-art on NTU-RGB+D w.r.t. models using the same input data, i.e., joint information. | [
"Regularization",
"Attention Modules",
"Output Functions",
"Stochastic Optimization",
"Subword Segmentation",
"Normalization",
"Feedforward Networks",
"Transformers",
"Attention Mechanisms",
"Skip Connections"
] | [
"Action Recognition",
"Action Recognition In Videos",
"Activity Recognition",
"Skeleton Based Action Recognition"
] | [
"Layer Normalization",
"Byte Pair Encoding",
"BPE",
"Softmax",
"Adam",
"Transformer",
"Multi-Head Attention",
"Residual Connection",
"Scaled Dot-Product Attention",
"Dropout",
"Label Smoothing",
"Dense Connections"
] | [
"NTU RGB+D",
"Kinetics-Skeleton dataset",
"NTU RGB+D 120"
] | [
"Accuracy (CS)",
"Accuracy (Cross-Subject)",
"Accuracy (CV)",
"Accuracy (Cross-Setup)",
"Accuracy"
] | Skeleton-based Action Recognition via Spatial and Temporal Transformer Networks |
Convolutional Neural Networks (CNN) have been regarded as a powerful class of
models for image recognition problems. Nevertheless, it is not trivial when
utilizing a CNN for learning spatio-temporal video representation. A few
studies have shown that performing 3D convolutions is a rewarding approach to
capture both spatial and temporal dimensions in videos. However, the
development of a very deep 3D CNN from scratch results in expensive
computational cost and memory demand. A valid question is why not recycle
off-the-shelf 2D networks for a 3D CNN. In this paper, we devise multiple
variants of bottleneck building blocks in a residual learning framework by
simulating $3\times3\times3$ convolutions with $1\times3\times3$ convolutional
filters on spatial domain (equivalent to 2D CNN) plus $3\times1\times1$
convolutions to construct temporal connections on adjacent feature maps in
time. Furthermore, we propose a new architecture, named Pseudo-3D Residual Net
(P3D ResNet), that exploits all the variants of blocks but composes each in
different placement of ResNet, following the philosophy that enhancing
structural diversity with going deep could improve the power of neural
networks. Our P3D ResNet achieves clear improvements on Sports-1M video
classification dataset against 3D CNN and frame-based 2D CNN by 5.3% and 1.8%,
respectively. We further examine the generalization performance of video
representation produced by our pre-trained P3D ResNet on five different
benchmarks and three different tasks, demonstrating superior performances over
several state-of-the-art techniques. | [
"Initialization",
"Convolutional Neural Networks",
"Activation Functions",
"Normalization",
"Convolutions",
"Pooling Operations",
"Skip Connections",
"Skip Connection Blocks"
] | [
"Action Recognition",
"Video Classification"
] | [
"ResNet",
"Average Pooling",
"Batch Normalization",
"Convolution",
"1x1 Convolution",
"ReLU",
"Residual Network",
"Residual Connection",
"Bottleneck Residual Block",
"Kaiming Initialization",
"Residual Block",
"Global Average Pooling",
"Rectified Linear Units",
"Max Pooling"
] | [
"Sports-1M",
"UCF101"
] | [
"Video hit@5",
"Video hit@1 ",
"3-fold Accuracy",
"Clip Hit@1"
] | Learning Spatio-Temporal Representation with Pseudo-3D Residual Networks |
While very deep neural networks have shown effectiveness for computer vision and text classification applications, how to increase the network depth of neural machine translation (NMT) models for better translation quality remains a challenging problem. Directly stacking more blocks to the NMT model results in no improvement and even reduces performance. In this work, we propose an effective two-stage approach with three specially designed components to construct deeper NMT models, which result in significant improvements over the strong Transformer baselines on WMT$14$ English$\to$German and English$\to$French translation tasks\footnote{Our code is available at \url{https://github.com/apeterswu/Depth_Growing_NMT}}. | [
"Regularization",
"Output Functions",
"Stochastic Optimization",
"Attention Modules",
"Activation Functions",
"Subword Segmentation",
"Normalization",
"Feedforward Networks",
"Transformers",
"Attention Mechanisms",
"Skip Connections"
] | [
"Machine Translation",
"Text Classification"
] | [
"Layer Normalization",
"Byte Pair Encoding",
"BPE",
"Softmax",
"Adam",
"Transformer",
"Multi-Head Attention",
"Rectified Linear Units",
"ReLU",
"Residual Connection",
"Label Smoothing",
"Dropout",
"Scaled Dot-Product Attention",
"Dense Connections"
] | [
"WMT2014 English-German",
"WMT2014 English-French"
] | [
"BLEU score"
] | Depth Growing for Neural Machine Translation |
Self-supervision provides effective representations for downstream tasks without requiring labels. However, existing approaches lag behind fully supervised training and are often not thought beneficial beyond obviating or reducing the need for annotations. We find that self-supervision can benefit robustness in a variety of ways, including robustness to adversarial examples, label corruption, and common input corruptions. Additionally, self-supervision greatly benefits out-of-distribution detection on difficult, near-distribution outliers, so much so that it exceeds the performance of fully supervised methods. These results demonstrate the promise of self-supervision for improving robustness and uncertainty estimation and establish these tasks as new axes of evaluation for future self-supervised learning research. | [
"Initialization",
"Convolutional Neural Networks",
"Activation Functions",
"Normalization",
"Convolutions",
"Pooling Operations",
"Skip Connections",
"Skip Connection Blocks"
] | [
"Anomaly Detection",
"Outlier Detection",
"Out-of-Distribution Detection",
"Self-Supervised Learning",
"Unsupervised Anomaly Detection"
] | [
"ResNet",
"Average Pooling",
"Batch Normalization",
"Convolution",
"1x1 Convolution",
"ReLU",
"Residual Network",
"Residual Connection",
"Bottleneck Residual Block",
"Kaiming Initialization",
"Residual Block",
"Global Average Pooling",
"Rectified Linear Units",
"Max Pooling"
] | [
"CIFAR-10",
"CIFAR-10 vs CIFAR-100",
"One-class CIFAR-10",
"One-class ImageNet-30"
] | [
"AUROC",
"AUPR",
"FPR95"
] | Using Self-Supervised Learning Can Improve Model Robustness and Uncertainty |
Exponential families are widely used in machine learning; they include many distributions in continuous and discrete domains (e.g., Gaussian, Dirichlet, Poisson, and categorical distributions via the softmax transformation). Distributions in each of these families have fixed support. In contrast, for finite domains, there has been recent work on sparse alternatives to softmax (e.g. sparsemax and alpha-entmax), which have varying support, being able to assign zero probability to irrelevant categories. This paper expands that work in two directions: first, we extend alpha-entmax to continuous domains, revealing a link with Tsallis statistics and deformed exponential families. Second, we introduce continuous-domain attention mechanisms, deriving efficient gradient backpropagation algorithms for alpha in {1,2}. Experiments on attention-based text classification, machine translation, and visual question answering illustrate the use of continuous attention in 1D and 2D, showing that it allows attending to time intervals and compact regions. | [
"Output Functions"
] | [
"Machine Translation",
"Question Answering",
"Text Classification",
"Visual Question Answering"
] | [
"Sparsemax",
"Softmax"
] | [
"VQA v2 test-std",
"VQA v2 test-dev"
] | [
"overall",
"Accuracy"
] | Sparse and Continuous Attention Mechanisms |
Transformer-based models are unable to process long sequences due to their self-attention operation, which scales quadratically with the sequence length. To address this limitation, we introduce the Longformer with an attention mechanism that scales linearly with sequence length, making it easy to process documents of thousands of tokens or longer. Longformer's attention mechanism is a drop-in replacement for the standard self-attention and combines a local windowed attention with a task motivated global attention. Following prior work on long-sequence transformers, we evaluate Longformer on character-level language modeling and achieve state-of-the-art results on text8 and enwik8. In contrast to most prior work, we also pretrain Longformer and finetune it on a variety of downstream tasks. Our pretrained Longformer consistently outperforms RoBERTa on long document tasks and sets new state-of-the-art results on WikiHop and TriviaQA. We finally introduce the Longformer-Encoder-Decoder (LED), a Longformer variant for supporting long document generative sequence-to-sequence tasks, and demonstrate its effectiveness on the arXiv summarization dataset. | [
"Regularization",
"Attention Modules",
"Learning Rate Schedules",
"Stochastic Optimization",
"Output Functions",
"Activation Functions",
"Subword Segmentation",
"Normalization",
"Feedforward Networks",
"Transformers",
"Attention Mechanisms",
"Skip Connections",
"Attention Patterns"
] | [
"Language Modelling"
] | [
"Weight Decay",
"WordPiece",
"Layer Normalization",
"Sliding Window Attention",
"Softmax",
"Multi-Head Attention",
"Attention Dropout",
"AdamW",
"Longformer",
"Linear Warmup With Linear Decay",
"Global and Sliding Window Attention",
"Dilated Sliding Window Attention",
"Residual Connection",
"Scaled Dot-Product Attention",
"Dropout",
"GELU",
"Dense Connections",
"Gaussian Linear Error Units"
] | [
"enwik8",
"Hutter Prize"
] | [
"Number of params",
"Bit per Character (BPC)"
] | Longformer: The Long-Document Transformer |
Recurrent neural networks have become ubiquitous in computing representations
of sequential data, especially textual data in natural language processing. In
particular, Bidirectional LSTMs are at the heart of several neural models
achieving state-of-the-art performance in a wide variety of tasks in NLP.
However, BiLSTMs are known to suffer from sequential bias - the contextual
representation of a token is heavily influenced by tokens close to it in a
sentence. We propose a general and effective improvement to the BiLSTM model
which encodes each suffix and prefix of a sequence of tokens in both forward
and reverse directions. We call our model Suffix Bidirectional LSTM or
SuBiLSTM. This introduces an alternate bias that favors long range
dependencies. We apply SuBiLSTMs to several tasks that require sentence
modeling. We demonstrate that using SuBiLSTM instead of a BiLSTM in existing
models leads to improvements in performance in learning general sentence
representations, text classification, textual entailment and paraphrase
detection. Using SuBiLSTM we achieve new state-of-the-art results for
fine-grained sentiment classification and question classification. | [
"Recurrent Neural Networks",
"Activation Functions",
"Bidirectional Recurrent Neural Networks"
] | [
"Natural Language Inference",
"Sentiment Analysis",
"Text Classification"
] | [
"Long Short-Term Memory",
"BiLSTM",
"Tanh Activation",
"Bidirectional LSTM",
"LSTM",
"Sigmoid Activation"
] | [
"MR",
"SST-2 Binary classification",
"CR",
"SST-5 Fine-grained classification"
] | [
"Accuracy"
] | Improved Sentence Modeling using Suffix Bidirectional LSTM |
In this paper we present a new computer vision task, named video instance segmentation. The goal of this new task is simultaneous detection, segmentation and tracking of instances in videos. In words, it is the first time that the image instance segmentation problem is extended to the video domain. To facilitate research on this new task, we propose a large-scale benchmark called YouTube-VIS, which consists of 2883 high-resolution YouTube videos, a 40-category label set and 131k high-quality instance masks. In addition, we propose a novel algorithm called MaskTrack R-CNN for this task. Our new method introduces a new tracking branch to Mask R-CNN to jointly perform the detection, segmentation and tracking tasks simultaneously. Finally, we evaluate the proposed method and several strong baselines on our new dataset. Experimental results clearly demonstrate the advantages of the proposed algorithm and reveal insight for future improvement. We believe the video instance segmentation task will motivate the community along the line of research for video understanding. | [
"Output Functions",
"RoI Feature Extractors",
"Convolutions",
"Graph Embeddings",
"Instance Segmentation Models"
] | [
"Instance Segmentation",
"Semantic Segmentation",
"Video Instance Segmentation",
"Video Understanding"
] | [
"LINE",
"Softmax",
"Convolution",
"RoIAlign",
"Mask R-CNN",
"Large-scale Information Network Embedding"
] | [
"YouTube-VIS validation"
] | [
"AR10",
"AR1",
"AP75",
"AP50",
"mask AP"
] | Video Instance Segmentation |
We present a new versatile building block for deep point cloud processing architectures. This building block combines the ideas of spatial transformers and multi-view CNNs with the efficiency of standard convolutional layers in two and three-dimensional dense grids. The new block operates via multiple parallel heads, whereas each head differentiably rasterizes feature representations of individual points into a low-dimensional space, and then uses dense convolution to propagate information across points. The results of the processing of individual heads are then combined together resulting in the update of point features. Using the new block, we build architectures for both discriminative (point cloud segmentation, point cloud classification) and generative (point cloud inpainting and image-based point cloud reconstruction) tasks. The resulting architectures invariably achieve state-of-the-art performance for these tasks, demonstrating the versatility and universality of the new block for point cloud processing. | [
"Convolutions"
] | [
"Semantic Segmentation"
] | [
"Convolution"
] | [
"S3DIS"
] | [
"Mean IoU"
] | Cloud Transformers |
Heterogeneous Face Recognition (HFR) is a challenging issue because of the large domain discrepancy and a lack of heterogeneous data. This paper considers HFR as a dual generation problem, and proposes a novel Dual Variational Generation (DVG) framework. It generates large-scale new paired heterogeneous images with the same identity from noise, for the sake of reducing the domain gap of HFR. Specifically, we first introduce a dual variational autoencoder to represent a joint distribution of paired heterogeneous images. Then, in order to ensure the identity consistency of the generated paired heterogeneous images, we impose a distribution alignment in the latent space and a pairwise identity preserving in the image space. Moreover, the HFR network reduces the domain discrepancy by constraining the pairwise feature distances between the generated paired heterogeneous images. Extensive experiments on four HFR databases show that our method can significantly improve state-of-the-art results. The related code is available at https://github.com/BradyFU/DVG. | [
"Generative Models"
] | [
"Face Recognition",
"Heterogeneous Face Recognition"
] | [
"AutoEncoder"
] | [
"BUAA-VisNir",
"Oulu-CASIA NIR-VIS",
"IIIT-D Viewed Sketch",
"CASIA NIR-VIS 2.0"
] | [
"TAR @ FAR=0.01",
"TAR @ FAR=0.001"
] | Dual Variational Generation for Low-Shot Heterogeneous Face Recognition |
Standard video frame interpolation methods first estimate optical flow
between input frames and then synthesize an intermediate frame guided by
motion. Recent approaches merge these two steps into a single convolution
process by convolving input frames with spatially adaptive kernels that account
for motion and re-sampling simultaneously. These methods require large kernels
to handle large motion, which limits the number of pixels whose kernels can be
estimated at once due to the large memory demand. To address this problem, this
paper formulates frame interpolation as local separable convolution over input
frames using pairs of 1D kernels. Compared to regular 2D kernels, the 1D
kernels require significantly fewer parameters to be estimated. Our method
develops a deep fully convolutional neural network that takes two input frames
and estimates pairs of 1D kernels for all pixels simultaneously. Since our
method is able to estimate kernels and synthesizes the whole video frame at
once, it allows for the incorporation of perceptual loss to train the neural
network to produce visually pleasing frames. This deep neural network is
trained end-to-end using widely available video data without any human
annotation. Both qualitative and quantitative experiments show that our method
provides a practical solution to high-quality video frame interpolation. | [
"Convolutions"
] | [
"Optical Flow Estimation",
"Video Frame Interpolation"
] | [
"Convolution"
] | [
"Middlebury",
"Vimeo90k"
] | [
"PSNR",
"Interpolation Error"
] | Video Frame Interpolation via Adaptive Separable Convolution |
We address the problem of Visual Question Answering (VQA), which requires
joint image and language understanding to answer a question about a given
photograph. Recent approaches have applied deep image captioning methods based
on convolutional-recurrent networks to this problem, but have failed to model
spatial inference. To remedy this, we propose a model we call the Spatial
Memory Network and apply it to the VQA task. Memory networks are recurrent
neural networks with an explicit attention mechanism that selects certain parts
of the information stored in memory. Our Spatial Memory Network stores neuron
activations from different spatial regions of the image in its memory, and uses
the question to choose relevant regions for computing the answer, a process of
which constitutes a single "hop" in the network. We propose a novel spatial
attention architecture that aligns words with image patches in the first hop,
and obtain improved results by adding a second attention hop which considers
the whole question to choose visual evidence based on the results of the first
hop. To better understand the inference process learned by the network, we
design synthetic questions that specifically require spatial inference and
visualize the attention weights. We evaluate our model on two published visual
question answering datasets, DAQUAR [1] and VQA [2], and obtain improved
results compared to a strong deep baseline model (iBOWIMG) which concatenates
image and question features to predict the answer [3]. | [
"Working Memory Models"
] | [
"Image Captioning",
"Question Answering",
"Visual Question Answering"
] | [
"Memory Network"
] | [
"COCO Visual Question Answering (VQA) real images 1.0 open ended"
] | [
"Percentage correct"
] | Ask, Attend and Answer: Exploring Question-Guided Spatial Attention for Visual Question Answering |
Neural text generation models are often autoregressive language models or seq2seq models. Neural autoregressive and seq2seq models that generate text by sampling words sequentially, with each word conditioned on the previous model, are state-of-the-art for several machine translation and summarization benchmarks. These benchmarks are often defined by validation perplexity even though this is not a direct measure of sample quality. Language models are typically trained via maximum likelihood and most often with teacher forcing. Teacher forcing is well-suited to optimizing perplexity but can result in poor sample quality because generating text requires conditioning on sequences of words that were never observed at training time. We propose to improve sample quality using Generative Adversarial Network (GANs), which explicitly train the generator to produce high quality samples and have shown a lot of success in image generation. GANs were originally to designed to output differentiable values, so discrete language generation is challenging for them. We introduce an actor-critic conditional GAN that fills in missing text conditioned on the surrounding context. We show qualitatively and quantitatively, evidence that this produces more realistic text samples compared to a maximum likelihood trained model. | [
"Recurrent Neural Networks",
"Activation Functions",
"Sequence To Sequence Models",
"Convolutions",
"Generative Models"
] | [
"Image Generation",
"Machine Translation",
"Multivariate Time Series Imputation",
"Text Generation"
] | [
"Generative Adversarial Network",
"Long Short-Term Memory",
"GAN",
"Convolution",
"Tanh Activation",
"Sequence to Sequence",
"LSTM",
"Seq2Seq",
"Sigmoid Activation"
] | [
"PEMS-SF"
] | [
"L2 Loss (10^-4)"
] | MaskGAN: Better Text Generation via Filling in the _______ |
Deep metric learning aims to learn a function mapping image pixels to embedding feature vectors that model the similarity between images. Two major applications of metric learning are content-based image retrieval and face verification. For the retrieval tasks, the majority of current state-of-the-art (SOTA) approaches are triplet-based non-parametric training. For the face verification tasks, however, recent SOTA approaches have adopted classification-based parametric training. In this paper, we look into the effectiveness of classification based approaches on image retrieval datasets. We evaluate on several standard retrieval datasets such as CAR-196, CUB-200-2011, Stanford Online Product, and In-Shop datasets for image retrieval and clustering, and establish that our classification-based approach is competitive across different feature dimensions and base feature networks. We further provide insights into the performance effects of subsampling classes for scalable classification-based training, and the effects of binarization, enabling efficient storage and computation for practical applications. | [
"Initialization",
"Convolutional Neural Networks",
"Activation Functions",
"Normalization",
"Convolutions",
"Pooling Operations",
"Skip Connections",
"Skip Connection Blocks"
] | [
"Binarization",
"Content-Based Image Retrieval",
"Face Verification",
"Image Retrieval",
"Metric Learning"
] | [
"ResNet",
"Average Pooling",
"Batch Normalization",
"Convolution",
"1x1 Convolution",
"ReLU",
"Residual Network",
"Residual Connection",
"Bottleneck Residual Block",
"Kaiming Initialization",
"Residual Block",
"Global Average Pooling",
"Rectified Linear Units",
"Max Pooling"
] | [
"CARS196",
"In-Shop",
"CUB-200-2011",
"SOP"
] | [
"R@1"
] | Classification is a Strong Baseline for Deep Metric Learning |
We present a new dataset for Visual Question Answering (VQA) on document images called DocVQA. The dataset consists of 50,000 questions defined on 12,000+ document images. Detailed analysis of the dataset in comparison with similar datasets for VQA and reading comprehension is presented. We report several baseline results by adopting existing VQA and reading comprehension models. Although the existing models perform reasonably well on certain types of questions, there is large performance gap compared to human performance (94.36% accuracy). The models need to improve specifically on questions where understanding structure of the document is crucial. The dataset, code and leaderboard are available at docvqa.org | [
"Language Models"
] | [
"Question Answering",
"Reading Comprehension",
"Visual Question Answering"
] | [
"BERT"
] | [
"DocVQA val",
"DocVQA test"
] | [
"Accuracy",
"ANLS"
] | DocVQA: A Dataset for VQA on Document Images |
Most methods for medical image segmentation use U-Net or its variants as they have been successful in most of the applications. After a detailed analysis of these "traditional" encoder-decoder based approaches, we observed that they perform poorly in detecting smaller structures and are unable to segment boundary regions precisely. This issue can be attributed to the increase in receptive field size as we go deeper into the encoder. The extra focus on learning high level features causes the U-Net based approaches to learn less information about low-level features which are crucial for detecting small structures. To overcome this issue, we propose using an overcomplete convolutional architecture where we project our input image into a higher dimension such that we constrain the receptive field from increasing in the deep layers of the network. We design a new architecture for image segmentation- KiU-Net which has two branches: (1) an overcomplete convolutional network Kite-Net which learns to capture fine details and accurate edges of the input, and (2) U-Net which learns high level features. Furthermore, we also propose KiU-Net 3D which is a 3D convolutional architecture for volumetric segmentation. We perform a detailed study of KiU-Net by performing experiments on five different datasets covering various image modalities like ultrasound (US), magnetic resonance imaging (MRI), computed tomography (CT), microscopic and fundus images. The proposed method achieves a better performance as compared to all the recent methods with an additional benefit of fewer parameters and faster convergence. Additionally, we also demonstrate that the extensions of KiU-Net based on residual blocks and dense blocks result in further performance improvements. The implementation of KiU-Net can be found here: https://github.com/jeya-maria-jose/KiU-Net-pytorch | [
"Semantic Segmentation Models",
"Activation Functions",
"Convolutions",
"Pooling Operations",
"Skip Connections"
] | [
"3D Medical Imaging Segmentation",
"Brain Tumor Segmentation",
"Computed Tomography (CT)",
"Liver Segmentation",
"Medical Image Segmentation",
"Semantic Segmentation",
"Ultrasound",
"Volumetric Medical Image Segmentation"
] | [
"U-Net",
"3D Convolution",
"Concatenated Skip Connection",
"Convolution",
"ReLU",
"Rectified Linear Units",
"Max Pooling"
] | [
"LiTS2017",
"RITE",
"Brain Anatomy US"
] | [
"Jaccard Index",
"IoU",
"Dice"
] | KiU-Net: Overcomplete Convolutional Architectures for Biomedical Image and Volumetric Segmentation |
As an important part of speech recognition technology, automatic speech keyword recognition has been intensively studied in recent years. Such technology becomes especially pivotal under situations with limited infrastructures and computational resources, such as voice command recognition in vehicles and robot interaction. At present, the mainstream methods in automatic speech keyword recognition are based on long short-term memory (LSTM) networks with attention mechanism. However, due to inevitable information losses for the LSTM layer caused during feature extraction, the calculated attention weights are biased. In this paper, a novel approach, namely Multi-layer Attention Mechanism, is proposed to handle the inaccurate attention weights problem. The key idea is that, in addition to the conventional attention mechanism, information of layers prior to feature extraction and LSTM are introduced into attention weights calculations. Therefore, the attention weights are more accurate because the overall model can have more precise and focused areas. We conduct a comprehensive comparison and analysis on the keyword spotting performances on convolution neural network, bi-directional LSTM cyclic neural network, and cyclic neural network with the proposed attention mechanism on Google Speech Command datasets V2 datasets. Experimental results indicate favorable results for the proposed method and demonstrate the validity of the proposed method. The proposed multi-layer attention methods can be useful for other researches related to object spotting. | [
"Convolutions",
"Activation Functions",
"Recurrent Neural Networks"
] | [
"Keyword Spotting",
"Speech Recognition"
] | [
"Long Short-Term Memory",
"Convolution",
"Tanh Activation",
"LSTM",
"Sigmoid Activation"
] | [
"Google Speech Commands"
] | [
"Google Speech Commands V2 20"
] | Multi-layer Attention Mechanism for Speech Keyword Recognition |
Bilinear models provide an appealing framework for mixing and merging
information in Visual Question Answering (VQA) tasks. They help to learn high
level associations between question meaning and visual concepts in the image,
but they suffer from huge dimensionality issues. We introduce MUTAN, a
multimodal tensor-based Tucker decomposition to efficiently parametrize
bilinear interactions between visual and textual representations. Additionally
to the Tucker framework, we design a low-rank matrix-based decomposition to
explicitly constrain the interaction rank. With MUTAN, we control the
complexity of the merging scheme while keeping nice interpretable fusion
relations. We show how our MUTAN model generalizes some of the latest VQA
architectures, providing state-of-the-art results. | [
"Distribution Approximation",
"Bijective Transformation"
] | [
"Visual Question Answering"
] | [
"Normalizing Flows",
"Affine Coupling"
] | [
"VQA v2 test-std",
"VQA v2 test-dev"
] | [
"overall",
"Accuracy"
] | MUTAN: Multimodal Tucker Fusion for Visual Question Answering |
Human parsing has received considerable interest due to its wide application
potentials. Nevertheless, it is still unclear how to develop an accurate human
parsing system in an efficient and elegant way. In this paper, we identify
several useful properties, including feature resolution, global context
information and edge details, and perform rigorous analyses to reveal how to
leverage them to benefit the human parsing task. The advantages of these useful
properties finally result in a simple yet effective Context Embedding with Edge
Perceiving (CE2P) framework for single human parsing. Our CE2P is end-to-end
trainable and can be easily adopted for conducting multiple human parsing.
Benefiting the superiority of CE2P, we achieved the 1st places on all three
human parsing benchmarks. Without any bells and whistles, we achieved 56.50\%
(mIoU), 45.31\% (mean $AP^r$) and 33.34\% ($AP^p_{0.5}$) in LIP, CIHP and MHP
v2.0, which outperform the state-of-the-arts more than 2.06\%, 3.81\% and
1.87\%, respectively. We hope our CE2P will serve as a solid baseline and help
ease future research in single/multiple human parsing. Code has been made
available at \url{https://github.com/liutinglt/CE2P}. | [
"Initialization",
"Convolutional Neural Networks",
"Activation Functions",
"Normalization",
"Convolutions",
"Pooling Operations",
"Skip Connections",
"Skip Connection Blocks"
] | [
"Human Parsing",
"Semantic Segmentation"
] | [
"ResNet",
"Average Pooling",
"Batch Normalization",
"Convolution",
"1x1 Convolution",
"ReLU",
"Residual Network",
"Residual Connection",
"Bottleneck Residual Block",
"Kaiming Initialization",
"Residual Block",
"Global Average Pooling",
"Rectified Linear Units",
"Max Pooling"
] | [
"LIP val"
] | [
"mIoU"
] | Devil in the Details: Towards Accurate Single and Multiple Human Parsing |
We tackle the problem of finding good architectures for multimodal classification problems. We propose a novel and generic search space that spans a large number of possible fusion architectures. In order to find an optimal architecture for a given dataset in the proposed search space, we leverage an efficient sequential model-based exploration approach that is tailored for the problem. We demonstrate the value of posing multimodal fusion as a neural architecture search problem by extensive experimentation on a toy dataset and two other real multimodal datasets. We discover fusion architectures that exhibit state-of-the-art performance for problems with different domain and dataset size, including the \ntu dataset, the largest multimodal action recognition dataset available.
| [
"Recurrent Neural Networks",
"Activation Functions",
"Output Functions"
] | [
"Action Recognition",
"Neural Architecture Search",
"Temporal Action Localization"
] | [
"Softmax",
"Long Short-Term Memory",
"Tanh Activation",
"LSTM",
"Sigmoid Activation"
] | [
"NTU RGB+D"
] | [
"Accuracy (CS)"
] | MFAS: Multimodal Fusion Architecture Search |
Normalization has become one of the most fundamental components in many deep neural networks for machine learning tasks while deep neural network has also been widely used in CTR estimation field. Among most of the proposed deep neural network models, few model utilize normalization approaches. Though some works such as Deep & Cross Network (DCN) and Neural Factorization Machine (NFM) use Batch Normalization in MLP part of the structure, there isn't work to thoroughly explore the effect of the normalization on the DNN ranking systems. In this paper, we conduct a systematic study on the effect of widely used normalization schemas by applying the various normalization approaches to both feature embedding and MLP part in DNN model. Extensive experiments are conduct on three real-world datasets and the experiment results demonstrate that the correct normalization significantly enhances model's performance. We also propose a new and effective normalization approaches based on LayerNorm named variance only LayerNorm(VO-LN) in this work. A normalization enhanced DNN model named NormDNN is also proposed based on the above-mentioned observation. As for the reason why normalization works for DNN models in CTR estimation, we find that the variance of normalization plays the main role and give an explanation in this work. | [
"Normalization"
] | [
"Click-Through Rate Prediction"
] | [
"Batch Normalization"
] | [
"Criteo",
"Android Malware Dataset"
] | [
"AUC"
] | Correct Normalization Matters: Understanding the Effect of Normalization On Deep Neural Network Models For Click-Through Rate Prediction |
The goal of graph representation learning is to embed each vertex in a graph
into a low-dimensional vector space. Existing graph representation learning
methods can be classified into two categories: generative models that learn the
underlying connectivity distribution in the graph, and discriminative models
that predict the probability of edge existence between a pair of vertices. In
this paper, we propose GraphGAN, an innovative graph representation learning
framework unifying above two classes of methods, in which the generative model
and discriminative model play a game-theoretical minimax game. Specifically,
for a given vertex, the generative model tries to fit its underlying true
connectivity distribution over all other vertices and produces "fake" samples
to fool the discriminative model, while the discriminative model tries to
detect whether the sampled vertex is from ground truth or generated by the
generative model. With the competition between these two models, both of them
can alternately and iteratively boost their performance. Moreover, when
considering the implementation of generative model, we propose a novel graph
softmax to overcome the limitations of traditional softmax function, which can
be proven satisfying desirable properties of normalization, graph structure
awareness, and computational efficiency. Through extensive experiments on
real-world datasets, we demonstrate that GraphGAN achieves substantial gains in
a variety of applications, including link prediction, node classification, and
recommendation, over state-of-the-art baselines. | [
"Output Functions"
] | [
"Graph Representation Learning",
"Link Prediction",
"Node Classification",
"Representation Learning"
] | [
"Softmax"
] | [
"BlogCatalog",
"Wikipedia"
] | [
"Macro-F1",
"Accuracy"
] | GraphGAN: Graph Representation Learning with Generative Adversarial Nets |
While deep learning has received a surge of interest in a variety of fields in recent years, major deep learning models barely use complex numbers. However, speech, signal and audio data are naturally complex-valued after Fourier Transform, and studies have shown a potentially richer representation of complex nets. In this paper, we propose a Complex Transformer, which incorporates the transformer model as a backbone for sequence modeling; we also develop attention and encoder-decoder network operating for complex input. The model achieves state-of-the-art performance on the MusicNet dataset and an In-phase Quadrature (IQ) signal dataset. | [
"Regularization",
"Output Functions",
"Stochastic Optimization",
"Attention Modules",
"Activation Functions",
"Subword Segmentation",
"Normalization",
"Feedforward Networks",
"Transformers",
"Attention Mechanisms",
"Skip Connections"
] | [
"Music Transcription"
] | [
"Layer Normalization",
"Byte Pair Encoding",
"BPE",
"Softmax",
"Adam",
"Transformer",
"Multi-Head Attention",
"Rectified Linear Units",
"ReLU",
"Residual Connection",
"Label Smoothing",
"Dropout",
"Scaled Dot-Product Attention",
"Dense Connections"
] | [
"MusicNet"
] | [
"APS",
"Number of params"
] | Complex Transformer: A Framework for Modeling Complex-Valued Sequence |
The ability to segment unknown objects in depth images has potential to
enhance robot skills in grasping and object tracking. Recent computer vision
research has demonstrated that Mask R-CNN can be trained to segment specific
categories of objects in RGB images when massive hand-labeled datasets are
available. As generating these datasets is time consuming, we instead train
with synthetic depth images. Many robots now use depth sensors, and recent
results suggest training on synthetic depth data can transfer successfully to
the real world. We present a method for automated dataset generation and
rapidly generate a synthetic training dataset of 50,000 depth images and
320,000 object masks using simulated heaps of 3D CAD models. We train a variant
of Mask R-CNN with domain randomization on the generated dataset to perform
category-agnostic instance segmentation without any hand-labeled data and we
evaluate the trained network, which we refer to as Synthetic Depth (SD) Mask
R-CNN, on a set of real, high-resolution depth images of challenging,
densely-cluttered bins containing objects with highly-varied geometry. SD Mask
R-CNN outperforms point cloud clustering baselines by an absolute 15% in
Average Precision and 20% in Average Recall on COCO benchmarks, and achieves
performance levels similar to a Mask R-CNN trained on a massive, hand-labeled
RGB dataset and fine-tuned on real images from the experimental setup. We
deploy the model in an instance-specific grasping pipeline to demonstrate its
usefulness in a robotics application. Code, the synthetic training dataset, and
supplementary material are available at https://bit.ly/2letCuE. | [
"Convolutions",
"RoI Feature Extractors",
"Output Functions",
"Instance Segmentation Models"
] | [
"Instance Segmentation",
"Object Tracking",
"Semantic Segmentation",
"Unseen Object Instance Segmentation"
] | [
"Mask R-CNN",
"Softmax",
"RoIAlign",
"Convolution"
] | [
"WISDOM"
] | [
"mAP @0.5:0.95"
] | Segmenting Unknown 3D Objects from Real Depth Images using Mask R-CNN Trained on Synthetic Data |
Batch Normalization (BN) uses mini-batch statistics to normalize the activations during training, introducing dependence between mini-batch elements. This dependency can hurt the performance if the mini-batch size is too small, or if the elements are correlated. Several alternatives, such as Batch Renormalization and Group Normalization (GN), have been proposed to address this issue. However, they either do not match the performance of BN for large batches, or still exhibit degradation in performance for smaller batches, or introduce artificial constraints on the model architecture. In this paper we propose the Filter Response Normalization (FRN) layer, a novel combination of a normalization and an activation function, that can be used as a replacement for other normalizations and activations. Our method operates on each activation channel of each batch element independently, eliminating the dependency on other batch elements. Our method outperforms BN and other alternatives in a variety of settings for all batch sizes. FRN layer performs $\approx 0.7-1.0\%$ better than BN on top-1 validation accuracy with large mini-batch sizes for Imagenet classification using InceptionV3 and ResnetV2-50 architectures. Further, it performs $>1\%$ better than GN on the same problem in the small mini-batch size regime. For object detection problem on COCO dataset, FRN layer outperforms all other methods by at least $0.3-0.5\%$ in all batch size regimes. | [
"Image Model Blocks",
"Initialization",
"Regularization",
"Convolutional Neural Networks",
"Learning Rate Schedules",
"Feature Extractors",
"Stochastic Optimization",
"Activation Functions",
"Output Functions",
"Miscellaneous Components",
"Loss Functions",
"Normalization",
"Convolutions",
"Feedforward Networks",
"Pooling Operations",
"Skip Connections",
"Object Detection Models",
"Skip Connection Blocks"
] | [
"Image Classification",
"Object Detection"
] | [
"Average Pooling",
"RMSProp",
"1x1 Convolution",
"Inception-v3",
"ResNet",
"VGG",
"Convolution",
"ReLU",
"Residual Connection",
"FPN",
"Dense Connections",
"Inception-v3 Module",
"Focal Loss",
"Batch Normalization",
"Residual Network",
"Label Smoothing",
"Kaiming Initialization",
"Filter Response Normalization",
"Softmax",
"Feature Pyramid Network",
"Auxiliary Classifier",
"Linear Warmup With Cosine Annealing",
"Bottleneck Residual Block",
"Dropout",
"RetinaNet",
"Residual Block",
"Global Average Pooling",
"Rectified Linear Units",
"Max Pooling"
] | [
"ImageNet"
] | [
"Top 5 Accuracy",
"Top 1 Accuracy"
] | Filter Response Normalization Layer: Eliminating Batch Dependence in the Training of Deep Neural Networks |
We present Deformable PV-RCNN, a high-performing point-cloud based 3D object detector. Currently, the proposal refinement methods used by the state-of-the-art two-stage detectors cannot adequately accommodate differing object scales, varying point-cloud density, part-deformation and clutter. We present a proposal refinement module inspired by 2D deformable convolution networks that can adaptively gather instance-specific features from locations where informative content exists. We also propose a simple context gating mechanism which allows the keypoints to select relevant context information for the refinement stage. We show state-of-the-art results on the KITTI dataset. | [
"Convolutions"
] | [
"3D Object Detection",
"Object Detection"
] | [
"Deformable Convolution",
"Convolution"
] | [
"KITTI Pedestrians Moderate val",
"KITTI Cyclists Moderate val",
"KITTI Cars Moderate val"
] | [
"AP"
] | Deformable PV-RCNN: Improving 3D Object Detection with Learned Deformations |
Lip-reading aims to recognize speech content from videos via visual analysis of speakers' lip movements. This is a challenging task due to the existence of homophemes-words which involve identical or highly similar lip movements, as well as diverse lip appearances and motion patterns among the speakers. To address these challenges, we propose a novel lip-reading model which captures not only the nuance between words but also styles of different speakers, by a multi-grained spatio-temporal modeling of the speaking process. Specifically, we first extract both frame-level fine-grained features and short-term medium-grained features by the visual front-end, which are then combined to obtain discriminative representations for words with similar phonemes. Next, a bidirectional ConvLSTM augmented with temporal attention aggregates spatio-temporal information in the entire input sequence, which is expected to be able to capture the coarse-gained patterns of each word and robust to various conditions in speaker identity, lighting conditions, and so on. By making full use of the information from different levels in a unified framework, the model is not only able to distinguish words with similar pronunciations, but also becomes robust to appearance changes. We evaluate our method on two challenging word-level lip-reading benchmarks and show the effectiveness of the proposed method, which also demonstrate the above claims. | [
"Convolutions",
"Activation Functions",
"Recurrent Neural Networks"
] | [
"Lipreading",
"Lip Reading"
] | [
"Tanh Activation",
"ConvLSTM",
"Sigmoid Activation",
"Convolution"
] | [
"Lip Reading in the Wild"
] | [
"Top-1 Accuracy"
] | Multi-Grained Spatio-temporal Modeling for Lip-reading |
BERT model has been successfully applied to open-domain QA tasks. However, previous work trains BERT by viewing passages corresponding to the same question as independent training instances, which may cause incomparable scores for answers from different passages. To tackle this issue, we propose a multi-passage BERT model to globally normalize answer scores across all passages of the same question, and this change enables our QA model find better answers by utilizing more passages. In addition, we find that splitting articles into passages with the length of 100 words by sliding window improves performance by 4%. By leveraging a passage ranker to select high-quality passages, multi-passage BERT gains additional 2%. Experiments on four standard benchmarks showed that our multi-passage BERT outperforms all state-of-the-art models on all benchmarks. In particular, on the OpenSQuAD dataset, our model gains 21.4% EM and 21.5% $F_1$ over all non-BERT models, and 5.8% EM and 6.5% $F_1$ over BERT-based models. | [
"Regularization",
"Output Functions",
"Learning Rate Schedules",
"Stochastic Optimization",
"Attention Modules",
"Activation Functions",
"Subword Segmentation",
"Normalization",
"Language Models",
"Feedforward Networks",
"Attention Mechanisms",
"Skip Connections"
] | [
"Open-Domain Question Answering",
"Question Answering"
] | [
"Weight Decay",
"WordPiece",
"Layer Normalization",
"Softmax",
"Adam",
"Multi-Head Attention",
"Attention Dropout",
"Linear Warmup With Linear Decay",
"Residual Connection",
"Scaled Dot-Product Attention",
"Dropout",
"BERT",
"GELU",
"Dense Connections",
"Gaussian Linear Error Units"
] | [
"SearchQA",
"Quasart-T"
] | [
"EM"
] | Multi-passage BERT: A Globally Normalized BERT Model for Open-domain Question Answering |
In this work, we address the problem of 3D human pose estimation from a
sequence of 2D human poses. Although the recent success of deep networks has
led many state-of-the-art methods for 3D pose estimation to train deep networks
end-to-end to predict from images directly, the top-performing approaches have
shown the effectiveness of dividing the task of 3D pose estimation into two
steps: using a state-of-the-art 2D pose estimator to estimate the 2D pose from
images and then mapping them into 3D space. They also showed that a
low-dimensional representation like 2D locations of a set of joints can be
discriminative enough to estimate 3D pose with high accuracy. However,
estimation of 3D pose for individual frames leads to temporally incoherent
estimates due to independent error in each frame causing jitter. Therefore, in
this work we utilize the temporal information across a sequence of 2D joint
locations to estimate a sequence of 3D poses. We designed a
sequence-to-sequence network composed of layer-normalized LSTM units with
shortcut connections connecting the input to the output on the decoder side and
imposed temporal smoothness constraint during training. We found that the
knowledge of temporal consistency improves the best reported result on
Human3.6M dataset by approximately $12.2\%$ and helps our network to recover
temporally consistent 3D poses over a sequence of images even when the 2D pose
detector fails. | [
"Recurrent Neural Networks",
"Activation Functions"
] | [
"3D Human Pose Estimation",
"3D Pose Estimation",
"Pose Estimation"
] | [
"Tanh Activation",
"Long Short-Term Memory",
"LSTM",
"Sigmoid Activation"
] | [
"Human3.6M"
] | [
"Average MPJPE (mm)"
] | Exploiting temporal information for 3D pose estimation |
We apply BERT to coreference resolution, achieving strong improvements on the OntoNotes (+3.9 F1) and GAP (+11.5 F1) benchmarks. A qualitative analysis of model predictions indicates that, compared to ELMo and BERT-base, BERT-large is particularly better at distinguishing between related but distinct entities (e.g., President and CEO). However, there is still room for improvement in modeling document-level context, conversations, and mention paraphrasing. Our code and models are publicly available. | [
"Regularization",
"Attention Modules",
"Learning Rate Schedules",
"Stochastic Optimization",
"Recurrent Neural Networks",
"Activation Functions",
"Output Functions",
"Subword Segmentation",
"Word Embeddings",
"Normalization",
"Language Models",
"Feedforward Networks",
"Attention Mechanisms",
"Skip Connections",
"Bidirectional Recurrent Neural Networks"
] | [
"Coreference Resolution"
] | [
"Weight Decay",
"Adam",
"Long Short-Term Memory",
"BiLSTM",
"Tanh Activation",
"Scaled Dot-Product Attention",
"Gaussian Linear Error Units",
"Bidirectional LSTM",
"Residual Connection",
"Dense Connections",
"ELMo",
"Layer Normalization",
"GELU",
"Sigmoid Activation",
"WordPiece",
"Softmax",
"Multi-Head Attention",
"Attention Dropout",
"Linear Warmup With Linear Decay",
"LSTM",
"Dropout",
"BERT"
] | [
"OntoNotes",
"CoNLL 2012"
] | [
"Avg F1",
"F1"
] | BERT for Coreference Resolution: Baselines and Analysis |
Dense video captioning is a task of localizing interesting events from an untrimmed video and producing textual description (captions) for each localized event. Most of the previous works in dense video captioning are solely based on visual information and completely ignore the audio track. However, audio, and speech, in particular, are vital cues for a human observer in understanding an environment. In this paper, we present a new dense video captioning approach that is able to utilize any number of modalities for event description. Specifically, we show how audio and speech modalities may improve a dense video captioning model. We apply automatic speech recognition (ASR) system to obtain a temporally aligned textual description of the speech (similar to subtitles) and treat it as a separate input alongside video frames and the corresponding audio track. We formulate the captioning task as a machine translation problem and utilize recently proposed Transformer architecture to convert multi-modal input data into textual descriptions. We demonstrate the performance of our model on ActivityNet Captions dataset. The ablation studies indicate a considerable contribution from audio and speech components suggesting that these modalities contain substantial complementary information to video frames. Furthermore, we provide an in-depth analysis of the ActivityNet Caption results by leveraging the category tags obtained from original YouTube videos. Code is publicly available: github.com/v-iashin/MDVC | [
"Regularization",
"Output Functions",
"Stochastic Optimization",
"Attention Modules",
"Activation Functions",
"Normalization",
"Feedforward Networks",
"Transformers",
"Attention Mechanisms",
"Skip Connections"
] | [
"Dense Video Captioning",
"Video Captioning"
] | [
"Layer Normalization",
"Transformer",
"Softmax",
"Adam",
"Multi-Head Attention",
"ReLU",
"Residual Connection",
"Label Smoothing",
"Dropout",
"Scaled Dot-Product Attention",
"Dense Connections",
"Rectified Linear Units"
] | [
"ActivityNet Captions"
] | [
"BLEU-3",
"BLEU-4",
"METEOR"
] | Multi-modal Dense Video Captioning |
In this paper, we propose a novel unsupervised clustering approach exploiting
the hidden information that is indirectly introduced through a pseudo
classification objective. Specifically, we randomly assign a pseudo
parent-class label to each observation which is then modified by applying the
domain specific transformation associated with the assigned label. Generated
pseudo observation-label pairs are subsequently used to train a neural network
with Auto-clustering Output Layer (ACOL) that introduces multiple softmax nodes
for each pseudo parent-class. Due to the unsupervised objective based on
Graph-based Activity Regularization (GAR) terms, softmax duplicates of each
parent-class are specialized as the hidden information captured through the
help of domain specific transformations is propagated during training.
Ultimately we obtain a k-means friendly latent representation. Furthermore, we
demonstrate how the chosen transformation type impacts performance and helps
propagate the latent information that is useful in revealing unknown clusters.
Our results show state-of-the-art performance for unsupervised clustering tasks
on MNIST, SVHN and USPS datasets, with the highest accuracies reported to date
in the literature. | [
"Output Functions"
] | [
"Unsupervised Image Classification"
] | [
"Softmax"
] | [
"MNIST",
"SVHN"
] | [
"Acc",
"# of clusters (k)",
"Accuracy"
] | Learning Latent Representations in Neural Networks for Clustering through Pseudo Supervision and Graph-based Activity Regularization |
We consider the problem of learning distributed representations for entities and relations of multi-relational data so as to predict missing links therein. Convolutional neural networks have recently shown their superiority for this problem, bringing increased model expressiveness while remaining parameter efficient. Despite the success, previous convolution designs fail to model full interactions between input entities and relations, which potentially limits the performance of link prediction. In this work we introduce ConvR, an adaptive convolutional network designed to maximize entity-relation interactions in a convolutional fashion. ConvR adaptively constructs convolution filters from relation representations, and applies these filters across entity representations to generate convolutional features. As such, ConvR enables rich interactions between entity and relation representations at diverse regions, and all the convolutional features generated will be able to capture such interactions. We evaluate ConvR on multiple benchmark datasets. Experimental results show that: (1) ConvR performs substantially better than competitive baselines in almost all the metrics and on all the datasets; (2) Compared with state-of-the-art convolutional models, ConvR is not only more effective but also more efficient. It offers a 7{\%} increase in MRR and a 6{\%} increase in Hits@10, while saving 12{\%} in parameter storage. | [
"Convolutions"
] | [
"Link Prediction",
"Relational Reasoning"
] | [
"Convolution"
] | [
" FB15k",
"WN18RR",
"WN18",
"FB15k-237"
] | [
"Hits@10",
"MRR",
"Hits@3",
"Hits@1"
] | Adaptive Convolution for Multi-Relational Learning |
Aspect-level sentiment classification aims to determine the sentiment polarity of a review sentence towards an opinion target. A sentence could contain multiple sentiment-target pairs; thus the main challenge of this task is to separate different opinion contexts for different targets. To this end, attention mechanism has played an important role in previous state-of-the-art neural models. The mechanism is able to capture the importance of each context word towards a target by modeling their semantic associations. We build upon this line of research and propose two novel approaches for improving the effectiveness of attention. First, we propose a method for target representation that better captures the semantic meaning of the opinion target. Second, we introduce an attention model that incorporates syntactic information into the attention mechanism. We experiment on attention-based LSTM (Long Short-Term Memory) models using the datasets from SemEval 2014, 2015, and 2016. The experimental results show that the conventional attention-based LSTM can be substantially improved by incorporating the two approaches. | [
"Recurrent Neural Networks",
"Activation Functions",
"Graph Embeddings"
] | [
"Aspect-Based Sentiment Analysis",
"Sentiment Analysis"
] | [
"LINE",
"Long Short-Term Memory",
"Tanh Activation",
"LSTM",
"Large-scale Information Network Embedding",
"Sigmoid Activation"
] | [
"SemEval 2014 Task 4 Sub Task 2"
] | [
"Laptop (Acc)",
"Restaurant (Acc)",
"Mean Acc (Restaurant + Laptop)"
] | Effective Attention Modeling for Aspect-Level Sentiment Classification |
Recently, ocular biometrics in unconstrained environments using images obtained at visible wavelength have gained the researchers' attention, especially with images captured by mobile devices. Periocular recognition has been demonstrated to be an alternative when the iris trait is not available due to occlusions or low image resolution. However, the periocular trait does not have the high uniqueness presented in the iris trait. Thus, the use of datasets containing many subjects is essential to assess biometric systems' capacity to extract discriminating information from the periocular region. Also, to address the within-class variability caused by lighting and attributes in the periocular region, it is of paramount importance to use datasets with images of the same subject captured in distinct sessions. As the datasets available in the literature do not present all these factors, in this work, we present a new periocular dataset containing samples from 1,122 subjects, acquired in 3 sessions by 196 different mobile devices. The images were captured under unconstrained environments with just a single instruction to the participants: to place their eyes on a region of interest. We also performed an extensive benchmark with several Convolutional Neural Network (CNN) architectures and models that have been employed in state-of-the-art approaches based on Multi-class Classification, Multitask Learning, Pairwise Filters Network, and Siamese Network. The results achieved in the closed- and open-world protocol, considering the identification and verification tasks, show that this area still needs research and development. | [
"Twin Networks"
] | [
"Face Recognition",
"Image Classification",
"Multi-class Classification"
] | [
"Siamese Network"
] | [
"UHDB31",
"Imbalanced CUB-200-2011"
] | [
"Rank-1",
"Accuracy"
] | UFPR-Periocular: A Periocular Dataset Collected by Mobile Devices in Unconstrained Scenarios |
We present a new framework for Patch Distribution Modeling, PaDiM, to concurrently detect and localize anomalies in images in a one-class learning setting. PaDiM makes use of a pretrained convolutional neural network (CNN) for patch embedding, and of multivariate Gaussian distributions to get a probabilistic representation of the normal class. It also exploits correlations between the different semantic levels of CNN to better localize anomalies. PaDiM outperforms current state-of-the-art approaches for both anomaly detection and localization on the MVTec AD and STC datasets. To match real-world visual industrial inspection, we extend the evaluation protocol to assess performance of anomaly localization algorithms on non-aligned dataset. The state-of-the-art performance and low complexity of PaDiM make it a good candidate for many industrial applications. | [
"Initialization",
"Regularization",
"Convolutional Neural Networks",
"Stochastic Optimization",
"Activation Functions",
"Normalization",
"Convolutions",
"Feedforward Networks",
"Pooling Operations",
"Skip Connections",
"Image Model Blocks",
"Image Models",
"Skip Connection Blocks"
] | [
"Anomaly Detection",
"Unsupervised Anomaly Detection"
] | [
"Depthwise Convolution",
"Average Pooling",
"EfficientNet",
"RMSProp",
"1x1 Convolution",
"ResNet",
"Convolution",
"ReLU",
"Residual Connection",
"Wide Residual Block",
"Dense Connections",
"Max Pooling",
"Swish",
"Batch Normalization",
"Residual Network",
"Squeeze-and-Excitation Block",
"Pointwise Convolution",
"Kaiming Initialization",
"Sigmoid Activation",
"Inverted Residual Block",
"Bottleneck Residual Block",
"Dropout",
"Depthwise Separable Convolution",
"Residual Block",
"Global Average Pooling",
"Rectified Linear Units",
"WideResNet"
] | [
"MVTec AD"
] | [
"Detection AUROC",
"Segmentation AUROC"
] | PaDiM: a Patch Distribution Modeling Framework for Anomaly Detection and Localization |
In this paper, we present a novel unsupervised video summarization model that
requires no manual annotation. The proposed model termed Cycle-SUM adopts a new
cycle-consistent adversarial LSTM architecture that can effectively maximize
the information preserving and compactness of the summary video. It consists of
a frame selector and a cycle-consistent learning based evaluator. The selector
is a bi-direction LSTM network that learns video representations that embed the
long-range relationships among video frames. The evaluator defines a learnable
information preserving metric between original video and summary video and
"supervises" the selector to identify the most informative frames to form the
summary video. In particular, the evaluator is composed of two generative
adversarial networks (GANs), in which the forward GAN is learned to reconstruct
original video from summary video while the backward GAN learns to invert the
processing. The consistency between the output of such cycle learning is
adopted as the information preserving metric for video summarization. We
demonstrate the close relation between mutual information maximization and such
cycle learning procedure. Experiments on two video summarization benchmark
datasets validate the state-of-the-art performance and superiority of the
Cycle-SUM model over previous baselines. | [
"Generative Models",
"Convolutions",
"Activation Functions",
"Recurrent Neural Networks"
] | [
"Unsupervised Video Summarization",
"Video Summarization"
] | [
"Generative Adversarial Network",
"Long Short-Term Memory",
"GAN",
"Convolution",
"Tanh Activation",
"LSTM",
"Sigmoid Activation"
] | [
"TvSum",
"SumMe"
] | [
"F1-score"
] | Cycle-SUM: Cycle-consistent Adversarial LSTM Networks for Unsupervised Video Summarization |
Semantic segmentation requires both rich spatial information and sizeable
receptive field. However, modern approaches usually compromise spatial
resolution to achieve real-time inference speed, which leads to poor
performance. In this paper, we address this dilemma with a novel Bilateral
Segmentation Network (BiSeNet). We first design a Spatial Path with a small
stride to preserve the spatial information and generate high-resolution
features. Meanwhile, a Context Path with a fast downsampling strategy is
employed to obtain sufficient receptive field. On top of the two paths, we
introduce a new Feature Fusion Module to combine features efficiently. The
proposed architecture makes a right balance between the speed and segmentation
performance on Cityscapes, CamVid, and COCO-Stuff datasets. Specifically, for a
2048x1024 input, we achieve 68.4% Mean IOU on the Cityscapes test dataset with
speed of 105 FPS on one NVIDIA Titan XP card, which is significantly faster
than the existing methods with comparable performance. | [
"Initialization",
"Convolutional Neural Networks",
"Activation Functions",
"Normalization",
"Convolutions",
"Pooling Operations",
"Skip Connections",
"Skip Connection Blocks"
] | [
"Real-Time Semantic Segmentation",
"Semantic Segmentation"
] | [
"ResNet",
"Average Pooling",
"Batch Normalization",
"Convolution",
"1x1 Convolution",
"ReLU",
"Residual Network",
"Residual Connection",
"Bottleneck Residual Block",
"Kaiming Initialization",
"Residual Block",
"Global Average Pooling",
"Rectified Linear Units",
"Max Pooling"
] | [
"CamVid",
"SkyScapes-Dense",
"Cityscapes test"
] | [
"Time (ms)",
"Mean IoU",
"mIoU",
"Mean IoU (class)",
"Frame (fps)"
] | BiSeNet: Bilateral Segmentation Network for Real-time Semantic Segmentation |
Modern methods mainly regard lane detection as a problem of pixel-wise segmentation, which is struggling to address the problem of challenging scenarios and speed. Inspired by human perception, the recognition of lanes under severe occlusion and extreme lighting conditions is mainly based on contextual and global information. Motivated by this observation, we propose a novel, simple, yet effective formulation aiming at extremely fast speed and challenging scenarios. Specifically, we treat the process of lane detection as a row-based selecting problem using global features. With the help of row-based selecting, our formulation could significantly reduce the computational cost. Using a large receptive field on global features, we could also handle the challenging scenarios. Moreover, based on the formulation, we also propose a structural loss to explicitly model the structure of lanes. Extensive experiments on two lane detection benchmark datasets show that our method could achieve the state-of-the-art performance in terms of both speed and accuracy. A light-weight version could even achieve 300+ frames per second with the same resolution, which is at least 4x faster than previous state-of-the-art methods. Our code will be made publicly available. | [
"Initialization",
"Convolutional Neural Networks",
"Learning Rate Schedules",
"Stochastic Optimization",
"Activation Functions",
"Normalization",
"Convolutions",
"Pooling Operations",
"Skip Connections",
"Skip Connection Blocks"
] | [
"Lane Detection"
] | [
"ResNet",
"Cosine Annealing",
"Average Pooling",
"Adam",
"Batch Normalization",
"Convolution",
"1x1 Convolution",
"ReLU",
"Residual Network",
"Residual Connection",
"Bottleneck Residual Block",
"Kaiming Initialization",
"Residual Block",
"Global Average Pooling",
"Rectified Linear Units",
"Max Pooling"
] | [
"CULane"
] | [
"F1 score"
] | Ultra Fast Structure-aware Deep Lane Detection |
Financial sentiment analysis is a challenging task due to the specialized language and lack of labeled data in that domain. General-purpose models are not effective enough because of the specialized language used in a financial context. We hypothesize that pre-trained language models can help with this problem because they require fewer labeled examples and they can be further trained on domain-specific corpora. We introduce FinBERT, a language model based on BERT, to tackle NLP tasks in the financial domain. Our results show improvement in every measured metric on current state-of-the-art results for two financial sentiment analysis datasets. We find that even with a smaller training set and fine-tuning only a part of the model, FinBERT outperforms state-of-the-art machine learning methods. | [
"Output Functions",
"Regularization",
"Stochastic Optimization",
"Attention Modules",
"Learning Rate Schedules",
"Activation Functions",
"Normalization",
"Subword Segmentation",
"Language Models",
"Feedforward Networks",
"Attention Mechanisms",
"Skip Connections"
] | [
"Language Modelling",
"Sentiment Analysis"
] | [
"Weight Decay",
"Layer Normalization",
"WordPiece",
"Softmax",
"Adam",
"Multi-Head Attention",
"Attention Dropout",
"Linear Warmup With Linear Decay",
"Residual Connection",
"Scaled Dot-Product Attention",
"Dropout",
"BERT",
"GELU",
"Dense Connections",
"Gaussian Linear Error Units"
] | [
"FiQA",
"Financial PhraseBank"
] | [
"MSE",
"R^2",
"F1 score",
"Accuracy"
] | FinBERT: Financial Sentiment Analysis with Pre-trained Language Models |
Highly expressive directed latent variable models, such as sigmoid belief
networks, are difficult to train on large datasets because exact inference in
them is intractable and none of the approximate inference methods that have
been applied to them scale well. We propose a fast non-iterative approximate
inference method that uses a feedforward network to implement efficient exact
sampling from the variational posterior. The model and this inference network
are trained jointly by maximizing a variational lower bound on the
log-likelihood. Although the naive estimator of the inference model gradient is
too high-variance to be useful, we make it practical by applying several
straightforward model-independent variance reduction techniques. Applying our
approach to training sigmoid belief networks and deep autoregressive networks,
we show that it outperforms the wake-sleep algorithm on MNIST and achieves
state-of-the-art results on the Reuters RCV1 document dataset. | [
"Feedforward Networks"
] | [
"Latent Variable Models",
"Variational Inference"
] | [
"Dense Connections",
"Feedforward Network"
] | [
"200k Short Texts for Humor Detection"
] | [
"10-20% Mask PSNR"
] | Neural Variational Inference and Learning in Belief Networks |
Keypoint detection is one of the most important pre-processing steps in tasks
such as face modeling, recognition and verification. In this paper, we present
an iterative method for Keypoint Estimation and Pose prediction of
unconstrained faces by Learning Efficient H-CNN Regressors (KEPLER) for
addressing the face alignment problem. Recent state of the art methods have
shown improvements in face keypoint detection by employing Convolution Neural
Networks (CNNs). Although a simple feed forward neural network can learn the
mapping between input and output spaces, it cannot learn the inherent
structural dependencies. We present a novel architecture called H-CNN
(Heatmap-CNN) which captures structured global and local features and thus
favors accurate keypoint detecion. HCNN is jointly trained on the visibility,
fiducials and 3D-pose of the face. As the iterations proceed, the error
decreases making the gradients small and thus requiring efficient training of
DCNNs to mitigate this. KEPLER performs global corrections in pose and
fiducials for the first four iterations followed by local corrections in the
subsequent stage. As a by-product, KEPLER also provides 3D pose (pitch, yaw and
roll) of the face accurately. In this paper, we show that without using any 3D
information, KEPLER outperforms state of the art methods for alignment on
challenging datasets such as AFW and AFLW. | [
"Convolutions"
] | [
"Face Alignment",
"Head Pose Estimation",
"Keypoint Detection",
"Pose Estimation",
"Pose Prediction"
] | [
"Convolution"
] | [
"BIWI"
] | [
"MAE (trained with other data)"
] | KEPLER: Keypoint and Pose Estimation of Unconstrained Faces by Learning Efficient H-CNN Regressors |
Exciting new work on the generalization bounds for neural networks (NN) given by Neyshabur et al. , Bartlett et al. closely depend on two parameter-depenedent quantities: the Lipschitz constant upper-bound and the stable rank (a softer version of the rank operator). This leads to an interesting question of whether controlling these quantities might improve the generalization behaviour of NNs. To this end, we propose stable rank normalization (SRN), a novel, optimal, and computationally efficient weight-normalization scheme which minimizes the stable rank of a linear operator. Surprisingly we find that SRN, inspite of being non-convex problem, can be shown to have a unique optimal solution. Moreover, we show that SRN allows control of the data-dependent empirical Lipschitz constant, which in contrast to the Lipschitz upper-bound, reflects the true behaviour of a model on a given dataset. We provide thorough analyses to show that SRN, when applied to the linear layers of a NN for classification, provides striking improvements-11.3% on the generalization gap compared to the standard NN along with significant reduction in memorization. When applied to the discriminator of GANs (called SRN-GAN) it improves Inception, FID, and Neural divergence scores on the CIFAR 10/100 and CelebA datasets, while learning mappings with low empirical Lipschitz constants. | [
"Regularization"
] | [
"Generalization Bounds",
"Image Generation"
] | [
"SRN",
"Stable Rank Normalization"
] | [
"CIFAR-10"
] | [
"Inception score",
"FID"
] | Stable Rank Normalization for Improved Generalization in Neural Networks and GANs |
Benefiting from its succinctness and robustness, skeleton-based action recognition has recently attracted much attention. Most existing methods utilize local networks (e.g., recurrent, convolutional, and graph convolutional networks) to extract spatio-temporal dynamics hierarchically. As a consequence, the local and non-local dependencies, which contain more details and semantics respectively, are asynchronously captured in different level of layers. Moreover, existing methods are limited to the spatio-temporal domain and ignore information in the frequency domain. To better extract synchronous detailed and semantic information from multi-domains, we propose a residual frequency attention (rFA) block to focus on discriminative patterns in the frequency domain, and a synchronous local and non-local (SLnL) block to simultaneously capture the details and semantics in the spatio-temporal domain. Besides, a soft-margin focal loss (SMFL) is proposed to optimize the learning whole process, which automatically conducts data selection and encourages intrinsic margins in classifiers. Our approach significantly outperforms other state-of-the-art methods on several large-scale datasets. | [
"Loss Functions"
] | [
"Action Recognition",
"Skeleton Based Action Recognition",
"Temporal Action Localization"
] | [
"Focal Loss"
] | [
"NTU RGB+D",
"Kinetics-Skeleton dataset"
] | [
"Accuracy (CS)",
"Accuracy (CV)",
"Accuracy"
] | Skeleton-Based Action Recognition with Synchronous Local and Non-local Spatio-temporal Learning and Frequency Attention |
We present some updates to YOLO! We made a bunch of little design changes to
make it better. We also trained this new network that's pretty swell. It's a
little bigger than last time but more accurate. It's still fast though, don't
worry. At 320x320 YOLOv3 runs in 22 ms at 28.2 mAP, as accurate as SSD but
three times faster. When we look at the old .5 IOU mAP detection metric YOLOv3
is quite good. It achieves 57.9 mAP@50 in 51 ms on a Titan X, compared to 57.5
mAP@50 in 198 ms by RetinaNet, similar performance but 3.8x faster. As always,
all the code is online at https://pjreddie.com/yolo/ | [
"Generalized Linear Models",
"Proposal Filtering",
"Output Functions",
"Feature Extractors",
"Convolutional Neural Networks",
"Loss Functions",
"Normalization",
"Convolutions",
"Clustering",
"Pooling Operations",
"Skip Connections",
"Object Detection Models"
] | [
"Object Detection",
"Real-Time Object Detection"
] | [
"Logistic Regression",
"k-Means Clustering",
"YOLOv3",
"Average Pooling",
"Focal Loss",
"Non Maximum Suppression",
"Feature Pyramid Network",
"Softmax",
"SSD",
"Convolution",
"Batch Normalization",
"1x1 Convolution",
"Residual Connection",
"FPN",
"RetinaNet",
"Darknet-53",
"Global Average Pooling"
] | [
"COCO",
"COCO test-dev"
] | [
"box AP",
"inference time (ms)",
"FPS",
"MAP"
] | YOLOv3: An Incremental Improvement |
Relation classification is an important NLP task to extract relations between entities. The state-of-the-art methods for relation classification are primarily based on Convolutional or Recurrent Neural Networks. Recently, the pre-trained BERT model achieves very successful results in many NLP classification / sequence labeling tasks. Relation classification differs from those tasks in that it relies on information of both the sentence and the two target entities. In this paper, we propose a model that both leverages the pre-trained BERT language model and incorporates information from the target entities to tackle the relation classification task. We locate the target entities and transfer the information through the pre-trained architecture and incorporate the corresponding encoding of the two entities. We achieve significant improvement over the state-of-the-art method on the SemEval-2010 task 8 relational dataset. | [
"Regularization",
"Output Functions",
"Learning Rate Schedules",
"Stochastic Optimization",
"Attention Modules",
"Activation Functions",
"Subword Segmentation",
"Normalization",
"Language Models",
"Feedforward Networks",
"Attention Mechanisms",
"Skip Connections"
] | [
"Language Modelling",
"Relation Classification",
"Relation Extraction"
] | [
"Weight Decay",
"WordPiece",
"Layer Normalization",
"Softmax",
"Adam",
"Multi-Head Attention",
"Attention Dropout",
"Linear Warmup With Linear Decay",
"Residual Connection",
"Scaled Dot-Product Attention",
"Dropout",
"BERT",
"GELU",
"Dense Connections",
"Gaussian Linear Error Units"
] | [
"TACRED",
"SemEval-2010 Task 8"
] | [
"F1"
] | Enriching Pre-trained Language Model with Entity Information for Relation Classification |
Many deep learning models, developed in recent years, reach higher ImageNet accuracy than ResNet50, with fewer or comparable FLOPS count. While FLOPs are often seen as a proxy for network efficiency, when measuring actual GPU training and inference throughput, vanilla ResNet50 is usually significantly faster than its recent competitors, offering better throughput-accuracy trade-off. In this work, we introduce a series of architecture modifications that aim to boost neural networks' accuracy, while retaining their GPU training and inference efficiency. We first demonstrate and discuss the bottlenecks induced by FLOPs-optimizations. We then suggest alternative designs that better utilize GPU structure and assets. Finally, we introduce a new family of GPU-dedicated models, called TResNet, which achieve better accuracy and efficiency than previous ConvNets. Using a TResNet model, with similar GPU throughput to ResNet50, we reach 80.8 top-1 accuracy on ImageNet. Our TResNet models also transfer well and achieve state-of-the-art accuracy on competitive single-label classification datasets such as Stanford cars (96.0%), CIFAR-10 (99.0%), CIFAR-100 (91.5%) and Oxford-Flowers (99.1%). They also perform well on multi-label classification and object detection tasks. Implementation is available at: https://github.com/mrT23/TResNet. | [
"Image Model Blocks",
"Image Data Augmentation",
"Regularization",
"Convolutional Neural Networks",
"Recurrent Neural Networks",
"Activation Functions",
"Normalization",
"Convolutions",
"Feedforward Networks",
"Pooling Operations",
"Skip Connections",
"Downsampling"
] | [
"Fine-Grained Image Classification",
"Image Classification",
"Multi-Label Classification",
"Object Detection"
] | [
"Weight Decay",
"Average Pooling",
"Cutout",
"Long Short-Term Memory",
"Tanh Activation",
"1x1 Convolution",
"Softplus",
"AutoAugment",
"Mish",
"Convolution",
"ReLU",
"Residual Connection",
"Leaky ReLU",
"Dense Connections",
"In-Place Activated Batch Normalization",
"Batch Normalization",
"Label Smoothing",
"ColorJitter",
"Squeeze-and-Excitation Block",
"InPlace-ABN",
"Sigmoid Activation",
"Color Jitter",
"Anti-Alias Downsampling",
"TResNet",
"LSTM",
"Global Average Pooling",
"Rectified Linear Units"
] | [
"CIFAR-100",
"Oxford 102 Flowers",
"CIFAR-10",
"Flowers-102",
"MS-COCO",
"Stanford Cars",
"ImageNet"
] | [
"Number of params",
"mAP",
"Top 1 Accuracy",
"Percentage correct",
"Accuracy"
] | TResNet: High Performance GPU-Dedicated Architecture |
Facial Expression Recognition (FER) is a challenging task that improves natural human-computer interaction. This paper focuses on automatic FER on a single in-the-wild (ITW) image. ITW images suffer real problems of pose, direction, and input resolution. In this study, we propose a pyramid with super-resolution (PSR) network architecture to solve the ITW FER task. We also introduce a prior distribution label smoothing (PDLS) loss function that applies the additional prior knowledge of the confusion about each expression in the FER task. Experiments on the three most popular ITW FER datasets showed that our approach outperforms all the state-of-the-art methods. | [
"Regularization"
] | [
"Facial Expression Recognition",
"Super-Resolution"
] | [
"Label Smoothing"
] | [
"RAF-DB",
"AffectNet",
"FERPlus"
] | [
"Overall Accuracy",
"Accuracy (8 emotion)",
"Accuracy (7 emotion)",
"Avg. Accuracy",
"Accuracy"
] | Pyramid With Super Resolution for In-the-Wild Facial Expression Recognition |
Training large deep neural networks on massive datasets is computationally very challenging. There has been recent surge in interest in using large batch stochastic optimization methods to tackle this issue. The most prominent algorithm in this line of research is LARS, which by employing layerwise adaptive learning rates trains ResNet on ImageNet in a few minutes. However, LARS performs poorly for attention models like BERT, indicating that its performance gains are not consistent across tasks. In this paper, we first study a principled layerwise adaptation strategy to accelerate training of deep neural networks using large mini-batches. Using this strategy, we develop a new layerwise adaptive large batch optimization technique called LAMB; we then provide convergence analysis of LAMB as well as LARS, showing convergence to a stationary point in general nonconvex settings. Our empirical results demonstrate the superior performance of LAMB across various tasks such as BERT and ResNet-50 training with very little hyperparameter tuning. In particular, for BERT training, our optimizer enables use of very large batch sizes of 32868 without any degradation of performance. By increasing the batch size to the memory limit of a TPUv3 Pod, BERT training time can be reduced from 3 days to just 76 minutes (Table 1). The LAMB implementation is available at https://github.com/tensorflow/addons/blob/master/tensorflow_addons/optimizers/lamb.py | [
"Convolutional Neural Networks",
"Normalization",
"Large Batch Optimization",
"Attention Mechanisms",
"Regularization",
"Attention Modules",
"Activation Functions",
"Subword Segmentation",
"Convolutions",
"Pooling Operations",
"Stochastic Optimization",
"Feedforward Networks",
"Graph Embeddings",
"Skip Connection Blocks",
"Initialization",
"Output Functions",
"Learning Rate Schedules",
"Language Models",
"Skip Connections"
] | [
"Question Answering",
"Stochastic Optimization"
] | [
"Weight Decay",
"LARS",
"Average Pooling",
"Adam",
"1x1 Convolution",
"Scaled Dot-Product Attention",
"Gaussian Linear Error Units",
"ResNet",
"Convolution",
"ReLU",
"Residual Connection",
"Large-scale Information Network Embedding",
"Dense Connections",
"LINE",
"Layer Normalization",
"Batch Normalization",
"Residual Network",
"GELU",
"Kaiming Initialization",
"LAMB",
"WordPiece",
"Softmax",
"Multi-Head Attention",
"Attention Dropout",
"Linear Warmup With Linear Decay",
"Bottleneck Residual Block",
"Dropout",
"BERT",
"Residual Block",
"Global Average Pooling",
"Rectified Linear Units",
"Max Pooling"
] | [
"SQuAD1.1 dev"
] | [
"F1"
] | Large Batch Optimization for Deep Learning: Training BERT in 76 minutes |
A stereotype is an over-generalized belief about a particular group of people, e.g., Asians are good at math or Asians are bad drivers. Such beliefs (biases) are known to hurt target groups. Since pretrained language models are trained on large real world data, they are known to capture stereotypical biases. In order to assess the adverse effects of these models, it is important to quantify the bias captured in them. Existing literature on quantifying bias evaluates pretrained language models on a small set of artificially constructed bias-assessing sentences. We present StereoSet, a large-scale natural dataset in English to measure stereotypical biases in four domains: gender, profession, race, and religion. We evaluate popular models like BERT, GPT-2, RoBERTa, and XLNet on our dataset and show that these models exhibit strong stereotypical biases. We also present a leaderboard with a hidden test set to track the bias of future language models at https://stereoset.mit.edu | [
"Regularization",
"Output Functions",
"Learning Rate Schedules",
"Stochastic Optimization",
"Attention Modules",
"Activation Functions",
"Subword Segmentation",
"Normalization",
"Language Models",
"Tokenizers",
"Attention Mechanisms",
"Feedforward Networks",
"Transformers",
"Fine-Tuning",
"Skip Connections"
] | [
"Bias Detection"
] | [
"Weight Decay",
"Cosine Annealing",
"Adam",
"Scaled Dot-Product Attention",
"SentencePiece",
"RoBERTa",
"Gaussian Linear Error Units",
"XLNet",
"Residual Connection",
"Dense Connections",
"Layer Normalization",
"Discriminative Fine-Tuning",
"GELU",
"GPT-2",
"WordPiece",
"Byte Pair Encoding",
"BPE",
"Softmax",
"Multi-Head Attention",
"Attention Dropout",
"Linear Warmup With Cosine Annealing",
"Linear Warmup With Linear Decay",
"Dropout",
"BERT"
] | [
"StereoSet"
] | [
"ICAT Score"
] | StereoSet: Measuring stereotypical bias in pretrained language models |
Diagnosing different retinal diseases from Spectral Domain Optical Coherence Tomography (SD-OCT) images is a challenging task. Different automated approaches such as image processing, machine learning and deep learning algorithms have been used for early detection and diagnosis of retinal diseases. Unfortunately, these are prone to error and computational inefficiency, which requires further intervention from human experts. In this paper, we propose a novel convolution neural network architecture to successfully distinguish between different degeneration of retinal layers and their underlying causes. The proposed novel architecture outperforms other classification models while addressing the issue of gradient explosion. Our approach reaches near perfect accuracy of 99.8% and 100% for two separately available Retinal SD-OCT data-set respectively. Additionally, our architecture predicts retinal diseases in real time while outperforming human diagnosticians. | [
"Convolutions"
] | [
"Retinal OCT Disease Classification"
] | [
"Convolution"
] | [
"Srinivasan2014",
"OCT2017"
] | [
"Acc",
"Sensitivity"
] | Optic-Net: A Novel Convolutional Neural Network for Diagnosis of Retinal Diseases from Optical Tomography Images |
We present the first massively distributed architecture for deep
reinforcement learning. This architecture uses four main components: parallel
actors that generate new behaviour; parallel learners that are trained from
stored experience; a distributed neural network to represent the value function
or behaviour policy; and a distributed store of experience. We used our
architecture to implement the Deep Q-Network algorithm (DQN). Our distributed
algorithm was applied to 49 games from Atari 2600 games from the Arcade
Learning Environment, using identical hyperparameters. Our performance
surpassed non-distributed DQN in 41 of the 49 games and also reduced the
wall-time required to achieve these results by an order of magnitude on most
games. | [
"Q-Learning Networks",
"Convolutions",
"Feedforward Networks",
"Off-Policy TD Control"
] | [
"Atari Games"
] | [
"Q-Learning",
"Convolution",
"DQN",
"Dense Connections",
"Deep Q-Network"
] | [
"Atari 2600 Amidar",
"Atari 2600 River Raid",
"Atari 2600 Beam Rider",
"Atari 2600 Video Pinball",
"Atari 2600 Demon Attack",
"Atari 2600 Enduro",
"Atari 2600 Alien",
"Atari 2600 Boxing",
"Atari 2600 Bank Heist",
"Atari 2600 Tutankham",
"Atari 2600 Time Pilot",
"Atari 2600 Space Invaders",
"Atari 2600 Assault",
"Atari 2600 Gravitar",
"Atari 2600 Ice Hockey",
"Atari 2600 Bowling",
"Atari 2600 Private Eye",
"Atari 2600 Asterix",
"Atari 2600 Breakout",
"Atari 2600 Name This Game",
"Atari 2600 Crazy Climber",
"Atari 2600 Pong",
"Atari 2600 Krull",
"Atari 2600 Freeway",
"Atari 2600 James Bond",
"Atari 2600 Robotank",
"Atari 2600 Kangaroo",
"Atari 2600 Venture",
"Atari 2600 Asteroids",
"Atari 2600 Fishing Derby",
"Atari 2600 Ms. Pacman",
"Atari 2600 Seaquest",
"Atari 2600 Tennis",
"Atari 2600 Zaxxon",
"Atari 2600 Frostbite",
"Atari 2600 Star Gunner",
"Atari 2600 Double Dunk",
"Atari 2600 Battle Zone",
"Atari 2600 Gopher",
"Atari 2600 Road Runner",
"Atari 2600 Atlantis",
"Atari 2600 Kung-Fu Master",
"Atari 2600 Chopper Command",
"Atari 2600 Up and Down",
"Atari 2600 Montezuma's Revenge",
"Atari 2600 Wizard of Wor",
"Atari 2600 Q*Bert",
"Atari 2600 Centipede",
"Atari 2600 HERO"
] | [
"Score"
] | Massively Parallel Methods for Deep Reinforcement Learning |
Emotion recognition in conversations is crucial for the development of empathetic machines. Present methods mostly ignore the role of inter-speaker dependency relations while classifying emotions in conversations. In this paper, we address recognizing utterance-level emotions in dyadic conversational videos. We propose a deep neural framework, termed Conversational Memory Network (CMN), which leverages contextual information from the conversation history. In particular, CMN uses multimodal approach comprising audio, visual and textual features with gated recurrent units to model past utterances of each speaker into memories. These memories are then merged using attention-based hops to capture inter-speaker dependencies. Experiments show a significant improvement of 3 − 4{\%} in accuracy over the state of the art. | [
"Working Memory Models"
] | [
"Emotion Recognition",
"Emotion Recognition in Conversation"
] | [
"Memory Network"
] | [
"IEMOCAP",
"SEMAINE"
] | [
"MAE (Arousal)",
"MAE (Power)",
"MAE (Valence)",
"MAE (Expectancy)",
"F1"
] | Conversational Memory Network for Emotion Recognition in Dyadic Dialogue Videos |
We develop a probabilistic interpretation of two-stage object detection. We show that this probabilistic interpretation motivates a number of common empirical training practices. It also suggests changes to two-stage detection pipelines. Specifically, the first stage should infer proper object-vs-background likelihoods, which should then inform the overall score of the detector. A standard region proposal network (RPN) cannot infer this likelihood sufficiently well, but many one-stage detectors can. We show how to build a probabilistic two-stage detector from any state-of-the-art one-stage detector. The resulting detectors are faster and more accurate than both their one- and two-stage precursors. Our detector achieves 56.4 mAP on COCO test-dev with single-scale testing, outperforming all published results. Using a lightweight backbone, our detector achieves 49.2 mAP on COCO at 33 fps on a Titan Xp, outperforming the popular YOLOv4 model. | [
"Feature Extractors",
"Convolutions",
"Pooling Operations"
] | [
"Object Detection",
"Region Proposal"
] | [
"Average Pooling",
"Feature Pyramid Network",
"Convolution",
"1x1 Convolution",
"FPN",
"Max Pooling"
] | [
"COCO test-dev"
] | [
"APM",
"box AP",
"AP75",
"APS",
"APL",
"AP50"
] | Probabilistic two-stage detection |
Hate speech and toxic comments are a common concern of social media platform users. Although these comments are, fortunately, the minority in these platforms, they are still capable of causing harm. Therefore, identifying these comments is an important task for studying and preventing the proliferation of toxicity in social media. Previous work in automatically detecting toxic comments focus mainly in English, with very few work in languages like Brazilian Portuguese. In this paper, we propose a new large-scale dataset for Brazilian Portuguese with tweets annotated as either toxic or non-toxic or in different types of toxicity. We present our dataset collection and annotation process, where we aimed to select candidates covering multiple demographic groups. State-of-the-art BERT models were able to achieve 76% macro-F1 score using monolingual data in the binary case. We also show that large-scale monolingual data is still needed to create more accurate models, despite recent advances in multilingual approaches. An error analysis and experiments with multi-label classification show the difficulty of classifying certain types of toxic comments that appear less frequently in our data and highlights the need to develop models that are aware of different categories of toxicity. | [
"Attention Modules",
"Regularization",
"Stochastic Optimization",
"Learning Rate Schedules",
"Output Functions",
"Activation Functions",
"Subword Segmentation",
"Normalization",
"Language Models",
"Feedforward Networks",
"Attention Mechanisms",
"Skip Connections"
] | [
"Hate Speech Detection",
"Multi-Label Classification",
"Text Classification"
] | [
"Weight Decay",
"Layer Normalization",
"WordPiece",
"Softmax",
"Adam",
"Multi-Head Attention",
"Attention Dropout",
"Linear Warmup With Linear Decay",
"Residual Connection",
"Scaled Dot-Product Attention",
"Dropout",
"BERT",
"GELU",
"Dense Connections",
"Gaussian Linear Error Units"
] | [
"ToLD-Br"
] | [
"F1-score"
] | Toxic Language Detection in Social Media for Brazilian Portuguese: New Dataset and Multilingual Analysis |
We propose Unicoder-VL, a universal encoder that aims to learn joint representations of vision and language in a pre-training manner. Borrow ideas from cross-lingual pre-trained models, such as XLM and Unicoder, both visual and linguistic contents are fed into a multi-layer Transformer for the cross-modal pre-training, where three pre-trained tasks are employed, including Masked Language Modeling (MLM), Masked Object Classification (MOC) and Visual-linguistic Matching (VLM). The first two tasks learn context-aware representations for input tokens based on linguistic and visual contents jointly. The last task tries to predict whether an image and a text describe each other. After pretraining on large-scale image-caption pairs, we transfer Unicoder-VL to caption-based image-text retrieval and visual commonsense reasoning, with just one additional output layer. We achieve state-of-the-art or comparable results on both two tasks and show the powerful ability of the cross-modal pre-training. | [
"Regularization",
"Output Functions",
"Stochastic Optimization",
"Attention Modules",
"Activation Functions",
"Subword Segmentation",
"Normalization",
"Feedforward Networks",
"Transformers",
"Attention Mechanisms",
"Skip Connections"
] | [
"Language Modelling",
"Object Classification",
"Text-Image Retrieval",
"Visual Commonsense Reasoning"
] | [
"Layer Normalization",
"Byte Pair Encoding",
"XLM",
"BPE",
"Adam",
"Softmax",
"Multi-Head Attention",
"Attention Dropout",
"Transformer",
"Rectified Linear Units",
"ReLU",
"Residual Connection",
"Label Smoothing",
"Dropout",
"Scaled Dot-Product Attention",
"GELU",
"Dense Connections",
"Gaussian Linear Error Units"
] | [
"COCO (image as query)"
] | [
"Recall@10"
] | Unicoder-VL: A Universal Encoder for Vision and Language by Cross-modal Pre-training |
Sentence matching is widely used in various natural language tasks such as
natural language inference, paraphrase identification, and question answering.
For these tasks, understanding logical and semantic relationship between two
sentences is required but it is yet challenging. Although attention mechanism
is useful to capture the semantic relationship and to properly align the
elements of two sentences, previous methods of attention mechanism simply use a
summation operation which does not retain original features enough. Inspired by
DenseNet, a densely connected convolutional network, we propose a
densely-connected co-attentive recurrent neural network, each layer of which
uses concatenated information of attentive features as well as hidden features
of all the preceding recurrent layers. It enables preserving the original and
the co-attentive feature information from the bottommost word embedding layer
to the uppermost recurrent layer. To alleviate the problem of an
ever-increasing size of feature vectors due to dense concatenation operations,
we also propose to use an autoencoder after dense concatenation. We evaluate
our proposed architecture on highly competitive benchmark datasets related to
sentence matching. Experimental results show that our architecture, which
retains recurrent and attentive features, achieves state-of-the-art
performances for most of the tasks. | [
"Generative Models"
] | [
"Natural Language Inference",
"Paraphrase Identification",
"Question Answering"
] | [
"AutoEncoder"
] | [
"SNLI"
] | [
"Parameters",
"% Train Accuracy",
"% Test Accuracy"
] | Semantic Sentence Matching with Densely-connected Recurrent and Co-attentive Information |
Although Convolutional neural networks (CNNs) are widely used for plant disease detection, they require a large number of training samples when dealing with wide variety of heterogeneous background. In this work, a CNN based dual phase method has been proposed which can work effectively on small rice grain disease dataset with heterogeneity. At the first phase, Faster RCNN method is applied for cropping out the significant portion (rice grain) from the image. This initial phase results in a secondary dataset of rice grains devoid of heterogeneous background. Disease classification is performed on such derived and simplified samples using CNN architecture. Comparison of the dual phase approach with straight forward application of CNN on the small grain dataset shows the effectiveness of the proposed method which provides a 5 fold cross validation accuracy of 88.07%. | [
"Output Functions",
"RoI Feature Extractors",
"Convolutions",
"Region Proposal",
"Object Detection Models"
] | [
"Rice Grain Disease Detection"
] | [
"RPN",
"Faster R-CNN",
"Softmax",
"RoIPool",
"Convolution",
"Region Proposal Network"
] | [
"Rice Grain Disease Dataset"
] | [
"mAP"
] | Rice grain disease identification using dual phase convolutional neural network-based system aimed at small dataset |
Neural Machine Translation (NMT) is an end-to-end learning approach for
automated translation, with the potential to overcome many of the weaknesses of
conventional phrase-based translation systems. Unfortunately, NMT systems are
known to be computationally expensive both in training and in translation
inference. Also, most NMT systems have difficulty with rare words. These issues
have hindered NMT's use in practical deployments and services, where both
accuracy and speed are essential. In this work, we present GNMT, Google's
Neural Machine Translation system, which attempts to address many of these
issues. Our model consists of a deep LSTM network with 8 encoder and 8 decoder
layers using attention and residual connections. To improve parallelism and
therefore decrease training time, our attention mechanism connects the bottom
layer of the decoder to the top layer of the encoder. To accelerate the final
translation speed, we employ low-precision arithmetic during inference
computations. To improve handling of rare words, we divide words into a limited
set of common sub-word units ("wordpieces") for both input and output. This
method provides a good balance between the flexibility of "character"-delimited
models and the efficiency of "word"-delimited models, naturally handles
translation of rare words, and ultimately improves the overall accuracy of the
system. Our beam search technique employs a length-normalization procedure and
uses a coverage penalty, which encourages generation of an output sentence that
is most likely to cover all the words in the source sentence. On the WMT'14
English-to-French and English-to-German benchmarks, GNMT achieves competitive
results to state-of-the-art. Using a human side-by-side evaluation on a set of
isolated simple sentences, it reduces translation errors by an average of 60%
compared to Google's phrase-based production system. | [
"Recurrent Neural Networks",
"Activation Functions",
"Subword Segmentation"
] | [
"Machine Translation"
] | [
"WordPiece",
"Long Short-Term Memory",
"Tanh Activation",
"LSTM",
"Sigmoid Activation"
] | [
"WMT2014 English-German",
"WMT2014 English-French"
] | [
"BLEU score"
] | Google's Neural Machine Translation System: Bridging the Gap between Human and Machine Translation |
Pre-training convolutional neural networks with weakly-supervised and self-supervised strategies is becoming increasingly popular for several computer vision tasks. However, due to the lack of strong discriminative signals, these learned representations may overfit to the pre-training objective (e.g., hashtag prediction) and not generalize well to downstream tasks. In this work, we present a simple strategy - ClusterFit (CF) to improve the robustness of the visual representations learned during pre-training. Given a dataset, we (a) cluster its features extracted from a pre-trained network using k-means and (b) re-train a new network from scratch on this dataset using cluster assignments as pseudo-labels. We empirically show that clustering helps reduce the pre-training task-specific information from the extracted features thereby minimizing overfitting to the same. Our approach is extensible to different pre-training frameworks -- weak- and self-supervised, modalities -- images and videos, and pre-training tasks -- object and action classification. Through extensive transfer learning experiments on 11 different target datasets of varied vocabularies and granularities, we show that ClusterFit significantly improves the representation quality compared to the state-of-the-art large-scale (millions / billions) weakly-supervised image and video models and self-supervised image models. | [
"Self-Supervised Learning"
] | [
"Action Classification",
"Action Classification ",
"Image Classification",
"Transfer Learning"
] | [
"ClusterFit"
] | [
"iNaturalist 2018"
] | [
"Top-1 Accuracy"
] | ClusterFit: Improving Generalization of Visual Representations |
We present graph attention networks (GATs), novel neural network
architectures that operate on graph-structured data, leveraging masked
self-attentional layers to address the shortcomings of prior methods based on
graph convolutions or their approximations. By stacking layers in which nodes
are able to attend over their neighborhoods' features, we enable (implicitly)
specifying different weights to different nodes in a neighborhood, without
requiring any kind of costly matrix operation (such as inversion) or depending
on knowing the graph structure upfront. In this way, we address several key
challenges of spectral-based graph neural networks simultaneously, and make our
model readily applicable to inductive as well as transductive problems. Our GAT
models have achieved or matched state-of-the-art results across four
established transductive and inductive graph benchmarks: the Cora, Citeseer and
Pubmed citation network datasets, as well as a protein-protein interaction
dataset (wherein test graphs remain unseen during training). | [
"Graph Models"
] | [
"Document Classification",
"Graph Classification",
"Graph Embedding",
"Graph Regression",
"Link Prediction",
"Node Classification",
"Skeleton Based Action Recognition"
] | [
"Graph Attention Network",
"GAT"
] | [
"PPI",
"CIFAR10 100k",
"Cora (0.5%)",
"CiteSeer with Public Split: fixed 20 nodes per class",
"PubMed with Public Split: fixed 20 nodes per class",
"Lipophilicity ",
"PubMed (0.1%)",
"J-HMBD Early Action",
"ZINC-500k",
"Cora (3%)",
"Brazil Air-Traffic",
"Europe Air-Traffic",
"CiteSeer (0.5%)",
"ZINC 100k",
"PubMed (0.03%)",
"Cora",
"PubMed (0.05%)",
"Wiki-Vote",
"PATTERN 100k",
"CiteSeer (1%)",
"Cora with Public Split: fixed 20 nodes per class",
"Flickr",
"Citeseer",
"USA Air-Traffic",
"Pubmed",
"Cora (1%)"
] | [
"Validation",
"Training Split",
"RMSE",
"MAE",
"Accuracy (%)",
"F1",
"Accuracy",
"10%"
] | Graph Attention Networks |
Named entity recognition is a challenging task that has traditionally
required large amounts of knowledge in the form of feature engineering and
lexicons to achieve high performance. In this paper, we present a novel neural
network architecture that automatically detects word- and character-level
features using a hybrid bidirectional LSTM and CNN architecture, eliminating
the need for most feature engineering. We also propose a novel method of
encoding partial lexicon matches in neural networks and compare it to existing
approaches. Extensive evaluation shows that, given only tokenized text and
publicly available word embeddings, our system is competitive on the CoNLL-2003
dataset and surpasses the previously reported state of the art performance on
the OntoNotes 5.0 dataset by 2.13 F1 points. By using two lexicons constructed
from publicly-available sources, we establish new state of the art performance
with an F1 score of 91.62 on CoNLL-2003 and 86.28 on OntoNotes, surpassing
systems that employ heavy feature engineering, proprietary lexicons, and rich
entity linking information. | [
"Convolutions",
"Activation Functions",
"Recurrent Neural Networks",
"Bidirectional Recurrent Neural Networks"
] | [
"Entity Linking",
"Feature Engineering",
"Named Entity Recognition",
"Word Embeddings"
] | [
"Long Short-Term Memory",
"BiLSTM",
"Convolution",
"Tanh Activation",
"CNN BiLSTM",
"Bidirectional LSTM",
"CNN Bidirectional LSTM",
"LSTM",
"Sigmoid Activation"
] | [
"Ontonotes v5 (English)",
"CoNLL 2003 (English)"
] | [
"F1"
] | Named Entity Recognition with Bidirectional LSTM-CNNs |
Normalizing flows, autoregressive models, variational autoencoders (VAEs), and deep energy-based models are among competing likelihood-based frameworks for deep generative learning. Among them, VAEs have the advantage of fast and tractable sampling and easy-to-access encoding networks. However, they are currently outperformed by other models such as normalizing flows and autoregressive models. While the majority of the research in VAEs is focused on the statistical challenges, we explore the orthogonal direction of carefully designing neural architectures for hierarchical VAEs. We propose Nouveau VAE (NVAE), a deep hierarchical VAE built for image generation using depth-wise separable convolutions and batch normalization. NVAE is equipped with a residual parameterization of Normal distributions and its training is stabilized by spectral regularization. We show that NVAE achieves state-of-the-art results among non-autoregressive likelihood-based models on the MNIST, CIFAR-10, CelebA 64, and CelebA HQ datasets and it provides a strong baseline on FFHQ. For example, on CIFAR-10, NVAE pushes the state-of-the-art from 2.98 to 2.91 bits per dimension, and it produces high-quality images on CelebA HQ. To the best of our knowledge, NVAE is the first successful VAE applied to natural images as large as 256$\times$256 pixels. The source code is available at https://github.com/NVlabs/NVAE . | [
"Learning Rate Schedules",
"Variational Optimization",
"Stochastic Optimization",
"Activation Functions",
"Distribution Approximation",
"Normalization",
"Convolutions",
"Feedforward Networks",
"Pooling Operations",
"Generative Models",
"Skip Connections",
"Image Model Blocks"
] | [
"Image Generation"
] | [
"Depthwise Convolution",
"Cosine Annealing",
"Average Pooling",
"Spectral Normalization",
"1x1 Convolution",
"Residual Normal Distribution",
"Weight Normalization",
"Variational Autoencoder",
"NVAE Encoder Residual Cell",
"Convolution",
"ReLU",
"Residual Connection",
"AdaMax",
"Dense Connections",
"NVAE Generative Residual Cell",
"Swish",
"Batch Normalization",
"VAE",
"NVAE",
"Pointwise Convolution",
"Squeeze-and-Excitation Block",
"Sigmoid Activation",
"Nouveau VAE",
"Normalizing Flows",
"Depthwise Separable Convolution",
"Rectified Linear Units"
] | [
"CelebA 256x256",
"ImageNet 32x32",
"FFHQ 256 x 256",
"CIFAR-10"
] | [
"bits/dimension",
"bpd"
] | NVAE: A Deep Hierarchical Variational Autoencoder |
Learning image representations without human supervision is an important and active research field. Several recent approaches have successfully leveraged the idea of making such a representation invariant under different types of perturbations, especially via contrastive-based instance discrimination training. Although effective visual representations should indeed exhibit such invariances, there are other important characteristics, such as encoding contextual reasoning skills, for which alternative reconstruction-based approaches might be better suited. With this in mind, we propose a teacher-student scheme to learn representations by training a convnet to reconstruct a bag-of-visual-words (BoW) representation of an image, given as input a perturbed version of that same image. Our strategy performs an online training of both the teacher network (whose role is to generate the BoW targets) and the student network (whose role is to learn representations), along with an online update of the visual-words vocabulary (used for the BoW targets). This idea effectively enables fully online BoW-guided unsupervised learning. Extensive experiments demonstrate the interest of our BoW-based strategy which surpasses previous state-of-the-art methods (including contrastive-based ones) in several applications. For instance, in downstream tasks such Pascal object detection, Pascal classification and Places205 classification, our method improves over all prior unsupervised approaches, thus establishing new state-of-the-art results that are also significantly better even than those of supervised pre-training. We provide the implementation code at https://github.com/valeoai/obow. | [
"Activation Functions",
"Normalization",
"Convolutions",
"Pooling Operations",
"Skip Connections",
"Skip Connection Blocks"
] | [
"Object Detection",
"Representation Learning",
"Self-Supervised Image Classification",
"Self-Supervised Learning",
"Semi-Supervised Image Classification",
"Unsupervised Pre-training",
"Unsupervised Representation Learning"
] | [
"Average Pooling",
"Batch Normalization",
"Convolution",
"1x1 Convolution",
"ReLU",
"Residual Connection",
"Bottleneck Residual Block",
"Residual Block",
"Global Average Pooling",
"Rectified Linear Units"
] | [
"ImageNet",
"ImageNet - 1% labeled data",
"ImageNet - 10% labeled data"
] | [
"Top 5 Accuracy",
"Number of Params",
"Top 1 Accuracy"
] | Online Bag-of-Visual-Words Generation for Unsupervised Representation Learning |
Over the past decade, Deep Convolutional Neural Networks have been widely adopted for medical image segmentation and shown to achieve adequate performance. However, due to the inherent inductive biases present in the convolutional architectures, they lack understanding of long-range dependencies in the image. Recently proposed Transformer-based architectures that leverage self-attention mechanism encode long-range dependencies and learn representations that are highly expressive. This motivates us to explore Transformer-based solutions and study the feasibility of using Transformer-based network architectures for medical image segmentation tasks. Majority of existing Transformer-based network architectures proposed for vision applications require large-scale datasets to train properly. However, compared to the datasets for vision applications, for medical imaging the number of data samples is relatively low, making it difficult to efficiently train transformers for medical applications. To this end, we propose a Gated Axial-Attention model which extends the existing architectures by introducing an additional control mechanism in the self-attention module. Furthermore, to train the model effectively on medical images, we propose a Local-Global training strategy (LoGo) which further improves the performance. Specifically, we operate on the whole image and patches to learn global and local features, respectively. The proposed Medical Transformer (MedT) is evaluated on three different medical image segmentation datasets and it is shown that it achieves better performance than the convolutional and other related transformer-based architectures. Code: https://github.com/jeya-maria-jose/Medical-Transformer | [
"Regularization",
"Output Functions",
"Stochastic Optimization",
"Attention Modules",
"Normalization",
"Subword Segmentation",
"Feedforward Networks",
"Transformers",
"Attention Mechanisms",
"Skip Connections"
] | [
"Medical Image Segmentation",
"Semantic Segmentation"
] | [
"Layer Normalization",
"Byte Pair Encoding",
"BPE",
"Softmax",
"Adam",
"Transformer",
"Multi-Head Attention",
"Residual Connection",
"Scaled Dot-Product Attention",
"Dropout",
"Label Smoothing",
"Dense Connections"
] | [
"MoNuSeg",
"GlaS",
"Brain US"
] | [
"IoU",
"F1"
] | Medical Transformer: Gated Axial-Attention for Medical Image Segmentation |
Graph Convolutional Networks (GCNs) have recently been shown to be quite successful in modeling graph-structured data. However, the primary focus has been on handling simple undirected graphs. Multi-relational graphs are a more general and prevalent form of graphs where each edge has a label and direction associated with it. Most of the existing approaches to handle such graphs suffer from over-parameterization and are restricted to learning representations of nodes only. In this paper, we propose CompGCN, a novel Graph Convolutional framework which jointly embeds both nodes and relations in a relational graph. CompGCN leverages a variety of entity-relation composition operations from Knowledge Graph Embedding techniques and scales with the number of relations. It also generalizes several of the existing multi-relational GCN methods. We evaluate our proposed method on multiple tasks such as node classification, link prediction, and graph classification, and achieve demonstrably superior results. We make the source code of CompGCN available to foster reproducible research. | [
"Graph Models"
] | [
"Graph Classification",
"Graph Embedding",
"Knowledge Graph Embedding",
"Link Prediction",
"Node Classification"
] | [
"Graph Convolutional Network",
"GCN"
] | [
"WN18RR",
"FB15k-237"
] | [
"Hits@3",
"Hits@1",
"MR",
"MRR",
"Hits@10"
] | Composition-based Multi-Relational Graph Convolutional Networks |
Along with studies on artificial intelligence technology, research is also being carried out actively in the field of natural language processing to understand and process people’s language, in other words, natural language. For computers to learn on their own, the skill of understanding natural language is very important. There are a wide variety of tasks involved in the field of natural language processing, but we would like to focus on the named entity registration and relation extraction task, which is considered to be the most important in understanding sentences. We propose DeNERT-KG, a model that can extract subject, object, and relationships, to grasp the meaning inherent in a sentence. Based on the BERT language model and Deep Q-Network, the named entity recognition (NER) model for extracting subject and object is established, and a knowledge graph is applied for relation extraction. Using the DeNERT-KG model, it is possible to extract the subject, type of subject, object, type of object, and relationship from a sentence, and verify this model through experiments. | [
"Regularization",
"Attention Modules",
"Learning Rate Schedules",
"Stochastic Optimization",
"Output Functions",
"Activation Functions",
"Subword Segmentation",
"Normalization",
"Language Models",
"Feedforward Networks",
"Attention Mechanisms",
"Skip Connections"
] | [
"Language Modelling",
"Named Entity Recognition",
"Relation Extraction"
] | [
"Weight Decay",
"WordPiece",
"Layer Normalization",
"Softmax",
"Adam",
"Multi-Head Attention",
"Attention Dropout",
"Linear Warmup With Linear Decay",
"Residual Connection",
"Scaled Dot-Product Attention",
"Dropout",
"BERT",
"GELU",
"Dense Connections",
"Gaussian Linear Error Units"
] | [
"TACRED"
] | [
"F1"
] | DeNERT-KG: Named Entity and Relation Extraction Model Using DQN, Knowledge Graph, and BERT |
Contexts play an important role in the saliency detection task. However,
given a context region, not all contextual information is helpful for the final
task. In this paper, we propose a novel pixel-wise contextual attention
network, i.e., the PiCANet, to learn to selectively attend to informative
context locations for each pixel. Specifically, for each pixel, it can generate
an attention map in which each attention weight corresponds to the contextual
relevance at each context location. An attended contextual feature can then be
constructed by selectively aggregating the contextual information. We formulate
the proposed PiCANet in both global and local forms to attend to global and
local contexts, respectively. Both models are fully differentiable and can be
embedded into CNNs for joint training. We also incorporate the proposed models
with the U-Net architecture to detect salient objects. Extensive experiments
show that the proposed PiCANets can consistently improve saliency detection
performance. The global and local PiCANets facilitate learning global contrast
and homogeneousness, respectively. As a result, our saliency model can detect
salient objects more accurately and uniformly, thus performing favorably
against the state-of-the-art methods. | [
"Semantic Segmentation Models",
"Activation Functions",
"Convolutions",
"Pooling Operations",
"Skip Connections"
] | [
"RGB Salient Object Detection",
"Saliency Detection"
] | [
"U-Net",
"Concatenated Skip Connection",
"Convolution",
"ReLU",
"Rectified Linear Units",
"Max Pooling"
] | [
"SOC",
"DUTS-TE"
] | [
"S-Measure",
"mean F-Measure",
"Average MAE",
"mean E-Measure",
"MAE",
"F-measure"
] | PiCANet: Learning Pixel-wise Contextual Attention for Saliency Detection |
For object detection, the two-stage approach (e.g., Faster R-CNN) has been
achieving the highest accuracy, whereas the one-stage approach (e.g., SSD) has
the advantage of high efficiency. To inherit the merits of both while
overcoming their disadvantages, in this paper, we propose a novel single-shot
based detector, called RefineDet, that achieves better accuracy than two-stage
methods and maintains comparable efficiency of one-stage methods. RefineDet
consists of two inter-connected modules, namely, the anchor refinement module
and the object detection module. Specifically, the former aims to (1) filter
out negative anchors to reduce search space for the classifier, and (2)
coarsely adjust the locations and sizes of anchors to provide better
initialization for the subsequent regressor. The latter module takes the
refined anchors as the input from the former to further improve the regression
and predict multi-class label. Meanwhile, we design a transfer connection block
to transfer the features in the anchor refinement module to predict locations,
sizes and class labels of objects in the object detection module. The
multi-task loss function enables us to train the whole network in an end-to-end
way. Extensive experiments on PASCAL VOC 2007, PASCAL VOC 2012, and MS COCO
demonstrate that RefineDet achieves state-of-the-art detection accuracy with
high efficiency. Code is available at https://github.com/sfzhang15/RefineDet | [
"Initialization",
"Regularization",
"Convolutional Neural Networks",
"Output Functions",
"Activation Functions",
"Normalization",
"Convolutions",
"Feedforward Networks",
"Pooling Operations",
"Skip Connections",
"Skip Connection Blocks"
] | [
"Object Detection",
"Regression"
] | [
"ResNet",
"VGG",
"Average Pooling",
"Softmax",
"Batch Normalization",
"Convolution",
"Rectified Linear Units",
"ReLU",
"1x1 Convolution",
"Residual Connection",
"Bottleneck Residual Block",
"Dropout",
"Residual Network",
"Kaiming Initialization",
"Residual Block",
"Global Average Pooling",
"Dense Connections",
"Max Pooling"
] | [
"PASCAL VOC 2007",
"COCO test-dev"
] | [
"APM",
"MAP",
"box AP",
"AP75",
"APS",
"APL",
"AP50"
] | Single-Shot Refinement Neural Network for Object Detection |
Scene text detection methods based on deep learning have achieved remarkable
results over the past years. However, due to the high diversity and complexity
of natural scenes, previous state-of-the-art text detection methods may still
produce a considerable amount of false positives, when applied to images
captured in real-world environments. To tackle this issue, mainly inspired by
Mask R-CNN, we propose in this paper an effective model for scene text
detection, which is based on Feature Pyramid Network (FPN) and instance
segmentation. We propose a supervised pyramid context network (SPCNET) to
precisely locate text regions while suppressing false positives. Benefited from
the guidance of semantic information and sharing FPN, SPCNET obtains
significantly enhanced performance while introducing marginal extra
computation. Experiments on standard datasets demonstrate that our SPCNET
clearly outperforms start-of-the-art methods. Specifically, it achieves an
F-measure of 92.1% on ICDAR2013, 87.2% on ICDAR2015, 74.1% on ICDAR2017 MLT and
82.9% on Total-Text. | [
"Feature Extractors",
"Convolutions"
] | [
"Instance Segmentation",
"Scene Text",
"Scene Text Detection",
"Semantic Segmentation"
] | [
"1x1 Convolution",
"Feature Pyramid Network",
"FPN",
"Convolution"
] | [
"ICDAR 2013",
"ICDAR 2017 MLT",
"ICDAR 2015",
"Total-Text"
] | [
"F-Measure",
"Recall",
"Precision"
] | Scene Text Detection with Supervised Pyramid Context Network |
Contrastive self-supervised learning (CSL) is an approach to learn useful representations by solving a pretext task that selects and compares anchor, negative and positive (APN) features from an unlabeled dataset. We present a conceptual framework that characterizes CSL approaches in five aspects (1) data augmentation pipeline, (2) encoder selection, (3) representation extraction, (4) similarity measure, and (5) loss function. We analyze three leading CSL approaches--AMDIM, CPC, and SimCLR--, and show that despite different motivations, they are special cases under this framework. We show the utility of our framework by designing Yet Another DIM (YADIM) which achieves competitive results on CIFAR-10, STL-10 and ImageNet, and is more robust to the choice of encoder and the representation extraction strategy. To support ongoing CSL research, we release the PyTorch implementation of this conceptual framework along with standardized implementations of AMDIM, CPC (V2), SimCLR, BYOL, Moco (V2) and YADIM. | [
"Self-Supervised Learning",
"Image Data Augmentation",
"Initialization",
"Convolutional Neural Networks",
"Activation Functions",
"Normalization",
"Loss Functions",
"Convolutions",
"Feedforward Networks",
"Pooling Operations",
"Skip Connections",
"Skip Connection Blocks"
] | [
"Data Augmentation",
"Image Classification",
"Self-Supervised Learning"
] | [
"InfoNCE",
"Average Pooling",
"1x1 Convolution",
"Normalized Temperature-scaled Cross Entropy Loss",
"ResNet",
"MoCo",
"Convolution",
"ReLU",
"SimCLR",
"Residual Connection",
"Dense Connections",
"Feedforward Network",
"Momentum Contrast",
"Random Resized Crop",
"Batch Normalization",
"Residual Network",
"ColorJitter",
"Kaiming Initialization",
"Color Jitter",
"NT-Xent",
"Random Gaussian Blur",
"Bottleneck Residual Block",
"Residual Block",
"Global Average Pooling",
"Rectified Linear Units",
"Max Pooling"
] | [
"STL-10"
] | [
"Percentage correct"
] | A Framework For Contrastive Self-Supervised Learning And Designing A New Approach |
Electrocardiography (ECG) signals are commonly used to diagnose various cardiac abnormalities. Recently, deep learning models showed initial success on modeling ECG data, however they are mostly black-box, thus lack interpretability needed for clinical usage. In this work, we propose MultIlevel kNowledge-guided Attention networks (MINA) that predict heart diseases from ECG signals with intuitive explanation aligned with medical knowledge. By extracting multilevel (beat-, rhythm- and frequency-level) domain knowledge features separately, MINA combines the medical knowledge and ECG data via a multilevel attention model, making the learned models highly interpretable. Our experiments showed MINA achieved PR-AUC 0.9436 (outperforming the best baseline by 5.51%) in real world ECG dataset. Finally, MINA also demonstrated robust performance and strong interpretability against signal distortion and noise contamination. | [
"Image Models"
] | [
"Electrocardiography (ECG)"
] | [
"Interpretability"
] | [
"PhysioNet Challenge 2017"
] | [
"ROC-AUC",
"PR-AUC",
"F1"
] | MINA: Multilevel Knowledge-Guided Attention for Modeling Electrocardiography Signals |
We present a novel deep neural network architecture for unsupervised subspace
clustering. This architecture is built upon deep auto-encoders, which
non-linearly map the input data into a latent space. Our key idea is to
introduce a novel self-expressive layer between the encoder and the decoder to
mimic the "self-expressiveness" property that has proven effective in
traditional subspace clustering. Being differentiable, our new self-expressive
layer provides a simple but effective way to learn pairwise affinities between
all data points through a standard back-propagation procedure. Being nonlinear,
our neural-network based method is able to cluster data points having complex
(often nonlinear) structures. We further propose pre-training and fine-tuning
strategies that let us effectively learn the parameters of our subspace
clustering networks. Our experiments show that the proposed method
significantly outperforms the state-of-the-art unsupervised subspace clustering
methods. | [
"Convolutions"
] | [
"Image Clustering"
] | [
"Convolution"
] | [
"Extended Yale-B"
] | [
"NMI",
"Accuracy"
] | Deep Subspace Clustering Networks |
Knowledge about the software used in scientific investigations is necessary for different reasons, including provenance of the results, measuring software impact to attribute developers, and bibliometric software citation analysis in general. Additionally, providing information about whether and how the software and the source code are available allows an assessment about the state and role of open source software in science in general. While such analyses can be done manually, large scale analyses require the application of automated methods of information extraction and linking. In this paper, we present SoftwareKG - a knowledge graph that contains information about software mentions from more than 51,000 scientific articles from the social sciences. A silver standard corpus, created by a distant and weak supervision approach, and a gold standard corpus, created by manual annotation, were used to train an LSTM based neural network to identify software mentions in scientific articles. The model achieves a recognition rate of .82 F-score in exact matches. As a result, we identified more than 133,000 software mentions. For entity disambiguation, we used the public domain knowledge base DBpedia. Furthermore, we linked the entities of the knowledge graph to other knowledge bases such as the Microsoft Academic Knowledge Graph, the Software Ontology, and Wikidata. Finally, we illustrate, how SoftwareKG can be used to assess the role of software in the social sciences. | [
"Recurrent Neural Networks",
"Activation Functions"
] | [
"Entity Disambiguation",
"Named Entity Recognition"
] | [
"Tanh Activation",
"Long Short-Term Memory",
"LSTM",
"Sigmoid Activation"
] | [
"SoSciSoCi"
] | [
"Precision",
"Recall",
"F1"
] | Investigating Software Usage in the Social Sciences: A Knowledge Graph Approach |
Deep clustering algorithms combine representation learning and clustering by jointly optimizing a clustering loss and a non-clustering loss. In such methods, a deep neural network is used for representation learning together with a clustering network. Instead of following this framework to improve clustering performance, we propose a simpler approach of optimizing the entanglement of the learned latent code representation of an autoencoder. We define entanglement as how close pairs of points from the same class or structure are, relative to pairs of points from different classes or structures. To measure the entanglement of data points, we use the soft nearest neighbor loss, and expand it by introducing an annealing temperature factor. Using our proposed approach, the test clustering accuracy was 96.2% on the MNIST dataset, 85.6% on the Fashion-MNIST dataset, and 79.2% on the EMNIST Balanced dataset, outperforming our baseline models. | [
"Image Representations"
] | [
"Deep Clustering",
"Image Clustering",
"Representation Learning"
] | [
"Annealing SNNL",
"Soft Nearest Neighbor Loss with Annealing Temperature"
] | [
"MNIST-test",
"Fashion-MNIST",
"EMNIST-Balanced"
] | [
"NMI",
"Accuracy"
] | Improving k-Means Clustering Performance with Disentangled Internal Representations |
Automatic detection and recognition of traffic signs plays a crucial role in
management of the traffic-sign inventory. It provides accurate and timely way
to manage traffic-sign inventory with a minimal human effort. In the computer
vision community the recognition and detection of traffic signs is a
well-researched problem. A vast majority of existing approaches perform well on
traffic signs needed for advanced drivers-assistance and autonomous systems.
However, this represents a relatively small number of all traffic signs (around
50 categories out of several hundred) and performance on the remaining set of
traffic signs, which are required to eliminate the manual labor in traffic-sign
inventory management, remains an open question. In this paper, we address the
issue of detecting and recognizing a large number of traffic-sign categories
suitable for automating traffic-sign inventory management. We adopt a
convolutional neural network (CNN) approach, the Mask R-CNN, to address the
full pipeline of detection and recognition with automatic end-to-end learning.
We propose several improvements that are evaluated on the detection of traffic
signs and result in an improved overall performance. This approach is applied
to detection of 200 traffic-sign categories represented in our novel dataset.
Results are reported on highly challenging traffic-sign categories that have
not yet been considered in previous works. We provide comprehensive analysis of
the deep learning method for the detection of traffic signs with large
intra-category appearance variation and show below 3% error rates with the
proposed approach, which is sufficient for deployment in practical applications
of traffic-sign inventory management. | [
"Initialization",
"Output Functions",
"Convolutional Neural Networks",
"RoI Feature Extractors",
"Activation Functions",
"Normalization",
"Convolutions",
"Pooling Operations",
"Instance Segmentation Models",
"Skip Connections",
"Skip Connection Blocks"
] | [
"Traffic Sign Detection",
"Traffic Sign Recognition"
] | [
"ResNet",
"Average Pooling",
"Softmax",
"Batch Normalization",
"Convolution",
"1x1 Convolution",
"ReLU",
"Residual Network",
"Residual Connection",
"RoIAlign",
"Bottleneck Residual Block",
"Mask R-CNN",
"Kaiming Initialization",
"Residual Block",
"Global Average Pooling",
"Rectified Linear Units",
"Max Pooling"
] | [
" Swedish traffic-sign dataset (STSD)",
"DFG traffic-sign dataset"
] | [
"[email protected]",
"mAP @0.5:0.95"
] | Deep Learning for Large-Scale Traffic-Sign Detection and Recognition |
We propose a simple, yet effective approach for spatiotemporal feature
learning using deep 3-dimensional convolutional networks (3D ConvNets) trained
on a large scale supervised video dataset. Our findings are three-fold: 1) 3D
ConvNets are more suitable for spatiotemporal feature learning compared to 2D
ConvNets; 2) A homogeneous architecture with small 3x3x3 convolution kernels in
all layers is among the best performing architectures for 3D ConvNets; and 3)
Our learned features, namely C3D (Convolutional 3D), with a simple linear
classifier outperform state-of-the-art methods on 4 different benchmarks and
are comparable with current best methods on the other 2 benchmarks. In
addition, the features are compact: achieving 52.8% accuracy on UCF101 dataset
with only 10 dimensions and also very efficient to compute due to the fast
inference of ConvNets. Finally, they are conceptually very simple and easy to
train and use. | [
"Convolutions"
] | [
"Action Recognition",
"Action Recognition In Videos"
] | [
"Convolution"
] | [
"Sports-1M",
"UCF101",
"HMDB-51"
] | [
"3-fold Accuracy",
"Video hit@1 ",
"Video hit@5",
"Average accuracy of 3 splits",
"Clip Hit@1"
] | Learning Spatiotemporal Features with 3D Convolutional Networks |
Existing domain adaptation focuses on transferring knowledge between domains with categorical indices (e.g., between datasets A and B). However, many tasks involve continuously indexed domains. For example, in medical applications, one often needs to transfer disease analysis and prediction across patients of different ages, where age acts as a continuous domain index. Such tasks are challenging for prior domain adaptation methods since they ignore the underlying relation among domains. In this paper, we propose the first method for continuously indexed domain adaptation. Our approach combines traditional adversarial adaptation with a novel discriminator that models the encoding-conditioned domain index distribution. Our theoretical analysis demonstrates the value of leveraging the domain index to generate invariant features across a continuous range of domains. Our empirical results show that our approach outperforms the state-of-the-art domain adaption methods on both synthetic and real-world medical datasets. | [
"Adversarial Training"
] | [
"Continuously Indexed Domain Adaptation",
"Domain Adaptation"
] | [
"CIDA",
"Continuously Indexed Domain Adaptation",
"PCIDA",
"Probabilistic Continuously Indexed Domain Adaptation"
] | [
"Sine",
"Indexed Rotating MNIST",
"Circle",
"Rotating MNIST"
] | [
"Accuracy (%)"
] | Continuously Indexed Domain Adaptation |
Accuracy and interpretability are two dominant features of successful
predictive models. Typically, a choice must be made in favor of complex black
box models such as recurrent neural networks (RNN) for accuracy versus less
accurate but more interpretable traditional models such as logistic regression.
This tradeoff poses challenges in medicine where both accuracy and
interpretability are important. We addressed this challenge by developing the
REverse Time AttentIoN model (RETAIN) for application to Electronic Health
Records (EHR) data. RETAIN achieves high accuracy while remaining clinically
interpretable and is based on a two-level neural attention model that detects
influential past visits and significant clinical variables within those visits
(e.g. key diagnoses). RETAIN mimics physician practice by attending the EHR
data in a reverse time order so that recent clinical visits are likely to
receive higher attention. RETAIN was tested on a large health system EHR
dataset with 14 million visits completed by 263K patients over an 8 year period
and demonstrated predictive accuracy and computational scalability comparable
to state-of-the-art methods such as RNN, and ease of interpretability
comparable to traditional models. | [
"Image Models"
] | [
"Disease Trajectory Forecasting",
"Regression"
] | [
"Interpretability"
] | [
"UK CF trust"
] | [
"AUC (Diabetes)",
"AUC (I. Obstruction)",
"AUC (K. Pneumonia)",
"AUC (Aspergillus)",
"AUC (E. Coli)",
"AUC (ABPA)",
"I. Obstruction"
] | RETAIN: An Interpretable Predictive Model for Healthcare using Reverse Time Attention Mechanism |
Neural architecture search (NAS) is a promising research direction that has the potential to replace expert-designed networks with learned, task-specific architectures. In this work, in order to help ground the empirical results in this field, we propose new NAS baselines that build off the following observations: (i) NAS is a specialized hyperparameter optimization problem; and (ii) random search is a competitive baseline for hyperparameter optimization. Leveraging these observations, we evaluate both random search with early-stopping and a novel random search with weight-sharing algorithm on two standard NAS benchmarks---PTB and CIFAR-10. Our results show that random search with early-stopping is a competitive NAS baseline, e.g., it performs at least as well as ENAS, a leading NAS method, on both benchmarks. Additionally, random search with weight-sharing outperforms random search with early-stopping, achieving a state-of-the-art NAS result on PTB and a highly competitive result on CIFAR-10. Finally, we explore the existing reproducibility issues of published NAS results. We note the lack of source material needed to exactly reproduce these results, and further discuss the robustness of published results given the various sources of variability in NAS experimental setups. Relatedly, we provide all information (code, random seeds, documentation) needed to exactly reproduce our results, and report our random search with weight-sharing results for each benchmark on multiple runs. | [
"Hyperparameter Search"
] | [
"Hyperparameter Optimization",
"Neural Architecture Search"
] | [
"Random Search"
] | [
"NAS-Bench-201, ImageNet-16-120",
"NAS-Bench-201, CIFAR-100",
"NAS-Bench-201, CIFAR-10"
] | [
"Search time (s)",
"Accuracy (Test)",
"Accuracy (Val)",
"Accuracy (val)"
] | Random Search and Reproducibility for Neural Architecture Search |
In recent years, supervised learning with convolutional networks (CNNs) has
seen huge adoption in computer vision applications. Comparatively, unsupervised
learning with CNNs has received less attention. In this work we hope to help
bridge the gap between the success of CNNs for supervised learning and
unsupervised learning. We introduce a class of CNNs called deep convolutional
generative adversarial networks (DCGANs), that have certain architectural
constraints, and demonstrate that they are a strong candidate for unsupervised
learning. Training on various image datasets, we show convincing evidence that
our deep convolutional adversarial pair learns a hierarchy of representations
from object parts to scenes in both the generator and discriminator.
Additionally, we use the learned features for novel tasks - demonstrating their
applicability as general image representations. | [
"Regularization",
"Stochastic Optimization",
"Activation Functions",
"Normalization",
"Convolutions",
"Generative Models",
"Non-Parametric Classification"
] | [
"Conditional Image Generation",
"Image Clustering",
"Representation Learning",
"Unsupervised Representation Learning"
] | [
"Weight Decay",
"Adam",
"Support Vector Machine",
"SVM",
"Batch Normalization",
"Convolution",
"ReLU",
"DCGAN",
"Deep Convolutional GAN",
"Leaky ReLU",
"Rectified Linear Units"
] | [
"Imagenet-dog-15",
"CIFAR-100",
"CIFAR-10",
"Tiny-ImageNet",
"ImageNet-10",
"STL-10",
"SVHN"
] | [
"Train set",
"Train Split",
"ARI",
"Percentage error",
"Backbone",
"Percentage correct",
"Train Set",
"NMI",
"Accuracy",
"Inception score"
] | Unsupervised Representation Learning with Deep Convolutional Generative Adversarial Networks |
Subsets and Splits