title
stringlengths 7
246
| abstract
stringlengths 6
3.31k
|
|---|---|
Bayesian filtering unifies adaptive and non-adaptive neural network
optimization methods
|
We formulate the problem of neural network optimization as Bayesian
filtering, where the observations are the backpropagated gradients. While
neural network optimization has previously been studied using natural gradient
methods which are closely related to Bayesian inference, they were unable to
recover standard optimizers such as Adam and RMSprop with a root-mean-square
gradient normalizer, instead getting a mean-square normalizer. To recover the
root-mean-square normalizer, we find it necessary to account for the temporal
dynamics of all the other parameters as they are geing optimized. The resulting
optimizer, AdaBayes, adaptively transitions between SGD-like and Adam-like
behaviour, automatically recovers AdamW, a state of the art variant of Adam
with decoupled weight decay, and has generalisation performance competitive
with SGD.
|
Understanding and Improving Interpolation in Autoencoders via an
Adversarial Regularizer
|
Autoencoders provide a powerful framework for learning compressed
representations by encoding all of the information needed to reconstruct a data
point in a latent code. In some cases, autoencoders can "interpolate": By
decoding the convex combination of the latent codes for two datapoints, the
autoencoder can produce an output which semantically mixes characteristics from
the datapoints. In this paper, we propose a regularization procedure which
encourages interpolated outputs to appear more realistic by fooling a critic
network which has been trained to recover the mixing coefficient from
interpolated data. We then develop a simple benchmark task where we can
quantitatively measure the extent to which various autoencoders can interpolate
and show that our regularizer dramatically improves interpolation in this
setting. We also demonstrate empirically that our regularizer produces latent
codes which are more effective on downstream tasks, suggesting a possible link
between interpolation abilities and learning useful representations.
|
Rearranging the Familiar: Testing Compositional Generalization in
Recurrent Networks
|
Systematic compositionality is the ability to recombine meaningful units with
regular and predictable outcomes, and it's seen as key to humans' capacity for
generalization in language. Recent work has studied systematic compositionality
in modern seq2seq models using generalization to novel navigation instructions
in a grounded environment as a probing tool, requiring models to quickly
bootstrap the meaning of new words. We extend this framework here to settings
where the model needs only to recombine well-trained functional words (such as
"around" and "right") in novel contexts. Our findings confirm and strengthen
the earlier ones: seq2seq models can be impressively good at generalizing to
novel combinations of previously-seen input, but only when they receive
extensive training on the specific pattern to be generalized (e.g.,
generalizing from many examples of "X around right" to "jump around right"),
while failing when generalization requires novel application of compositional
rules (e.g., inferring the meaning of "around right" from those of "right" and
"around").
|
Partial recovery bounds for clustering with the relaxed $K$means
|
We investigate the clustering performances of the relaxed $K$means in the
setting of sub-Gaussian Mixture Model (sGMM) and Stochastic Block Model (SBM).
After identifying the appropriate signal-to-noise ratio (SNR), we prove that
the misclassification error decay exponentially fast with respect to this SNR.
These partial recovery bounds for the relaxed $K$means improve upon results
currently known in the sGMM setting. In the SBM setting, applying the relaxed
$K$means SDP allows to handle general connection probabilities whereas other
SDPs investigated in the literature are restricted to the assortative case
(where within group probabilities are larger than between group probabilities).
Again, this partial recovery bound complements the state-of-the-art results.
All together, these results put forward the versatility of the relaxed
$K$means.
|
Compositional GAN: Learning Image-Conditional Binary Composition
|
Generative Adversarial Networks (GANs) can produce images of remarkable
complexity and realism but are generally structured to sample from a single
latent source ignoring the explicit spatial interaction between multiple
entities that could be present in a scene. Capturing such complex interactions
between different objects in the world, including their relative scaling,
spatial layout, occlusion, or viewpoint transformation is a challenging
problem. In this work, we propose a novel self-consistent
Composition-by-Decomposition (CoDe) network to compose a pair of objects. Given
object images from two distinct distributions, our model can generate a
realistic composite image from their joint distribution following the texture
and shape of the input objects. We evaluate our approach through qualitative
experiments and user evaluations. Our results indicate that the learned model
captures potential interactions between the two object domains, and generates
realistic composed scenes at test time.
|
Fully Convolutional Pixel Adaptive Image Denoiser
|
We propose a new image denoising algorithm, dubbed as Fully Convolutional
Adaptive Image DEnoiser (FC-AIDE), that can learn from an offline supervised
training set with a fully convolutional neural network as well as adaptively
fine-tune the supervised model for each given noisy image. We significantly
extend the framework of the recently proposed Neural AIDE, which formulates the
denoiser to be context-based pixelwise mappings and utilizes the unbiased
estimator of MSE for such denoisers. The two main contributions we make are; 1)
implementing a novel fully convolutional architecture that boosts the base
supervised model, and 2) introducing regularization methods for the adaptive
fine-tuning such that a stronger and more robust adaptivity can be attained. As
a result, FC-AIDE is shown to possess many desirable features; it outperforms
the recent CNN-based state-of-the-art denoisers on all of the benchmark
datasets we tested, and gets particularly strong for various challenging
scenarios, e.g., with mismatched image/noise characteristics or with scarce
supervised training data. The source code of our algorithm is available at
https://github.com/csm9493/FC-AIDE-Keras.
|
Doubly Stochastic Adversarial Autoencoder
|
Any autoencoder network can be turned into a generative model by imposing an
arbitrary prior distribution on its hidden code vector. Variational Autoencoder
(VAE) [2] uses a KL divergence penalty to impose the prior, whereas Adversarial
Autoencoder (AAE) [1] uses {\it generative adversarial networks} GAN [3]. GAN
trades the complexities of {\it sampling} algorithms with the complexities of
{\it searching} Nash equilibrium in minimax games. Such minimax architectures
get trained with the help of data examples and gradients flowing through a
generator and an adversary. A straightforward modification of AAE is to replace
the adversary with the maximum mean discrepancy (MMD) test [4-5]. This
replacement leads to a new type of probabilistic autoencoder, which is also
discussed in our paper. We propose a novel probabilistic autoencoder in which
the adversary of AAE is replaced with a space of {\it stochastic} functions.
This replacement introduces a new source of randomness, which can be considered
as a continuous control for encouraging {\it explorations}. This prevents the
adversary from fitting too closely to the generator and therefore leads to a
more diverse set of generated samples.
|
Unsupervised Metric Learning in Presence of Missing Data
|
For many machine learning tasks, the input data lie on a low-dimensional
manifold embedded in a high dimensional space and, because of this
high-dimensional structure, most algorithms are inefficient. The typical
solution is to reduce the dimension of the input data using standard dimension
reduction algorithms such as ISOMAP, LAPLACIAN EIGENMAPS or LLES. This
approach, however, does not always work in practice as these algorithms require
that we have somewhat ideal data. Unfortunately, most data sets either have
missing entries or unacceptably noisy values. That is, real data are far from
ideal and we cannot use these algorithms directly. In this paper, we focus on
the case when we have missing data. Some techniques, such as matrix completion,
can be used to fill in missing data but these methods do not capture the
non-linear structure of the manifold. Here, we present a new algorithm
MR-MISSING that extends these previous algorithms and can be used to compute
low dimensional representation on data sets with missing entries. We
demonstrate the effectiveness of our algorithm by running three different
experiments. We visually verify the effectiveness of our algorithm on synthetic
manifolds, we numerically compare our projections against those computed by
first filling in data using nlPCA and mDRUR on the MNIST data set, and we also
show that we can do classification on MNIST with missing data. We also provide
a theoretical guarantee for MR-MISSING under some simplifying assumptions.
|
Adaptive Variational Particle Filtering in Non-stationary Environments
|
Online convex optimization is a sequential prediction framework with the goal
to track and adapt to the environment through evaluating proper convex loss
functions. We study efficient particle filtering methods from the perspective
of such a framework.
We formulate an efficient particle filtering methods for the non-stationary
environment by making connections with the online mirror descent algorithm
which is known to be a universal online convex optimization algorithm.
As a result of this connection, our proposed particle filtering algorithm
proves to achieve optimal particle efficiency.
|
Approximate Collapsed Gibbs Clustering with Expectation Propagation
|
We develop a framework for approximating collapsed Gibbs sampling in
generative latent variable cluster models. Collapsed Gibbs is a popular MCMC
method, which integrates out variables in the posterior to improve mixing.
Unfortunately for many complex models, integrating out these variables is
either analytically or computationally intractable. We efficiently approximate
the necessary collapsed Gibbs integrals by borrowing ideas from expectation
propagation. We present two case studies where exact collapsed Gibbs sampling
is intractable: mixtures of Student-t's and time series clustering. Our
experiments on real and synthetic data show that our approximate sampler
enables a runtime-accuracy tradeoff in sampling these types of models,
providing results with competitive accuracy much more rapidly than the naive
Gibbs samplers one would otherwise rely on in these scenarios.
|
Tsallis-INF: An Optimal Algorithm for Stochastic and Adversarial Bandits
|
We derive an algorithm that achieves the optimal (within constants)
pseudo-regret in both adversarial and stochastic multi-armed bandits without
prior knowledge of the regime and time horizon. The algorithm is based on
online mirror descent (OMD) with Tsallis entropy regularization with power
$\alpha=1/2$ and reduced-variance loss estimators. More generally, we define an
adversarial regime with a self-bounding constraint, which includes stochastic
regime, stochastically constrained adversarial regime (Wei and Luo), and
stochastic regime with adversarial corruptions (Lykouris et al.) as special
cases, and show that the algorithm achieves logarithmic regret guarantee in
this regime and all of its special cases simultaneously with the adversarial
regret guarantee.} The algorithm also achieves adversarial and stochastic
optimality in the utility-based dueling bandit setting. We provide empirical
evaluation of the algorithm demonstrating that it significantly outperforms
UCB1 and EXP3 in stochastic environments. We also provide examples of
adversarial environments, where UCB1 and Thompson Sampling exhibit almost
linear regret, whereas our algorithm suffers only logarithmic regret. To the
best of our knowledge, this is the first example demonstrating vulnerability of
Thompson Sampling in adversarial environments. Last, but not least, we present
a general stochastic analysis and a general adversarial analysis of OMD
algorithms with Tsallis entropy regularization for $\alpha\in[0,1]$ and explain
the reason why $\alpha=1/2$ works best.
|
Rapid Time Series Prediction with a Hardware-Based Reservoir Computer
|
Reservoir computing is a neural network approach for processing
time-dependent signals that has seen rapid development in recent years.
Physical implementations of the technique using optical reservoirs have
demonstrated remarkable accuracy and processing speed at benchmark tasks.
However, these approaches require an electronic output layer to maintain high
performance, which limits their use in tasks such as time-series prediction,
where the output is fed back into the reservoir. We present here a reservoir
computing scheme that has rapid processing speed both by the reservoir and the
output layer. The reservoir is realized by an autonomous, time-delay, Boolean
network configured on a field-programmable gate array. We investigate the
dynamical properties of the network and observe the fading memory property that
is critical for successful reservoir computing. We demonstrate the utility of
the technique by training a reservoir to learn the short- and long-term
behavior of a chaotic system. We find accuracy comparable to state-of-the-art
software approaches of similar network size, but with a superior real-time
prediction rate up to 160 MHz.
|
Automatically Designing CNN Architectures for Medical Image Segmentation
|
Deep neural network architectures have traditionally been designed and
explored with human expertise in a long-lasting trial-and-error process. This
process requires huge amount of time, expertise, and resources. To address this
tedious problem, we propose a novel algorithm to optimally find hyperparameters
of a deep network architecture automatically. We specifically focus on
designing neural architectures for medical image segmentation task. Our
proposed method is based on a policy gradient reinforcement learning for which
the reward function is assigned a segmentation evaluation utility (i.e., dice
index). We show the efficacy of the proposed method with its low computational
cost in comparison with the state-of-the-art medical image segmentation
networks. We also present a new architecture design, a densely connected
encoder-decoder CNN, as a strong baseline architecture to apply the proposed
hyperparameter search algorithm. We apply the proposed algorithm to each layer
of the baseline architectures. As an application, we train the proposed system
on cine cardiac MR images from Automated Cardiac Diagnosis Challenge (ACDC)
MICCAI 2017. Starting from a baseline segmentation architecture, the resulting
network architecture obtains the state-of-the-art results in accuracy without
performing any trial-and-error based architecture design approaches or close
supervision of the hyperparameters changes.
|
Hierarchical Reinforcement Learning for Zero-shot Generalization with
Subtask Dependencies
|
We introduce a new RL problem where the agent is required to generalize to a
previously-unseen environment characterized by a subtask graph which describes
a set of subtasks and their dependencies. Unlike existing hierarchical
multitask RL approaches that explicitly describe what the agent should do at a
high level, our problem only describes properties of subtasks and relationships
among them, which requires the agent to perform complex reasoning to find the
optimal subtask to execute. To solve this problem, we propose a neural subtask
graph solver (NSGS) which encodes the subtask graph using a recursive neural
network embedding. To overcome the difficulty of training, we propose a novel
non-parametric gradient-based policy, graph reward propagation, to pre-train
our NSGS agent and further finetune it through actor-critic method. The
experimental results on two 2D visual domains show that our agent can perform
complex reasoning to find a near-optimal way of executing the subtask graph and
generalize well to the unseen subtask graphs. In addition, we compare our agent
with a Monte-Carlo tree search (MCTS) method showing that our method is much
more efficient than MCTS, and the performance of NSGS can be further improved
by combining it with MCTS.
|
Efficient Probabilistic Inference in the Quest for Physics Beyond the
Standard Model
|
We present a novel probabilistic programming framework that couples directly
to existing large-scale simulators through a cross-platform probabilistic
execution protocol, which allows general-purpose inference engines to record
and control random number draws within simulators in a language-agnostic way.
The execution of existing simulators as probabilistic programs enables highly
interpretable posterior inference in the structured model defined by the
simulator code base. We demonstrate the technique in particle physics, on a
scientifically accurate simulation of the tau lepton decay, which is a key
ingredient in establishing the properties of the Higgs boson. Inference
efficiency is achieved via inference compilation where a deep recurrent neural
network is trained to parameterize proposal distributions and control the
stochastic simulator in a sequential importance sampling scheme, at a fraction
of the computational cost of a Markov chain Monte Carlo baseline.
|
Brain Tumor Segmentation and Tractographic Feature Extraction from
Structural MR Images for Overall Survival Prediction
|
This paper introduces a novel methodology to integrate human brain
connectomics and parcellation for brain tumor segmentation and survival
prediction. For segmentation, we utilize an existing brain parcellation atlas
in the MNI152 1mm space and map this parcellation to each individual subject
data. We use deep neural network architectures together with hard negative
mining to achieve the final voxel level classification. For survival
prediction, we present a new method for combining features from connectomics
data, brain parcellation information, and the brain tumor mask. We leverage the
average connectome information from the Human Connectome Project and map each
subject brain volume onto this common connectome space. From this, we compute
tractographic features that describe potential neural disruptions due to the
brain tumor. These features are then used to predict the overall survival of
the subjects. The main novelty in the proposed methods is the use of normalized
brain parcellation data and tractography data from the human connectome project
for analyzing MR images for segmentation and survival prediction. Experimental
results are reported on the BraTS2018 dataset.
|
Learning Representations for Soft Skill Matching
|
Employers actively look for talents having not only specific hard skills but
also various soft skills. To analyze the soft skill demands on the job market,
it is important to be able to detect soft skill phrases from job advertisements
automatically. However, a naive matching of soft skill phrases can lead to
false positive matches when a soft skill phrase, such as friendly, is used to
describe a company, a team, or another entity, rather than a desired candidate.
In this paper, we propose a phrase-matching-based approach which
differentiates between soft skill phrases referring to a candidate vs.
something else. The disambiguation is formulated as a binary text
classification problem where the prediction is made for the potential soft
skill based on the context where it occurs. To inform the model about the soft
skill for which the prediction is made, we develop several approaches,
including soft skill masking and soft skill tagging.
We compare several neural network based approaches, including CNN, LSTM and
Hierarchical Attention Model. The proposed tagging-based input representation
using LSTM achieved the highest recall of 83.92% on the job dataset when fixing
a precision to 95%.
|
Twitter Sentiment Analysis System
|
Social media is increasingly used by humans to express their feelings and
opinions in the form of short text messages. Detecting sentiments in the text
has a wide range of applications including identifying anxiety or depression of
individuals and measuring well-being or mood of a community. Sentiments can be
expressed in many ways that can be seen such as facial expression and gestures,
speech and by written text. Sentiment Analysis in text documents is essentially
a content-based classification problem involving concepts from the domains of
Natural Language Processing as well as Machine Learning. In this paper,
sentiment recognition based on textual data and the techniques used in
sentiment analysis are discussed.
|
Learning the effect of latent variables in Gaussian Graphical models
with unobserved variables
|
The edge structure of the graph defining an undirected graphical model
describes precisely the structure of dependence between the variables in the
graph. In many applications, the dependence structure is unknown and it is
desirable to learn it from data, often because it is a preliminary step to be
able to ascertain causal effects. This problem, known as structure learning, is
hard in general, but for Gaussian graphical models it is slightly easier
because the structure of the graph is given by the sparsity pattern of the
precision matrix of the joint distribution, and because independence coincides
with decorrelation. A major difficulty too often ignored in structure learning
is the fact that if some variables are not observed, the marginal dependence
graph over the observed variables will possibly be significantly more complex
and no longer reflect the direct dependencies that are potentially associated
with causal effects. In this work, we consider a family of latent variable
Gaussian graphical models in which the graph of the joint distribution between
observed and unobserved variables is sparse, and the unobserved variables are
conditionally independent given the others. Prior work was able to recover the
connectivity between observed variables, but could only identify the subspace
spanned by unobserved variables, whereas we propose a convex optimization
formulation based on structured matrix sparsity to estimate the complete
connectivity of the complete graph including unobserved variables, given the
knowledge of the number of missing variables, and a priori knowledge of their
level of connectivity. Our formulation is supported by a theoretical result of
identifiability of the latent dependence structure for sparse graphs in the
infinite data limit. We propose an algorithm leveraging recent active set
methods, which performs well in the experiments on synthetic data.
|
Physical Adversarial Examples for Object Detectors
|
Deep neural networks (DNNs) are vulnerable to adversarial
examples-maliciously crafted inputs that cause DNNs to make incorrect
predictions. Recent work has shown that these attacks generalize to the
physical domain, to create perturbations on physical objects that fool image
classifiers under a variety of real-world conditions. Such attacks pose a risk
to deep learning models used in safety-critical cyber-physical systems. In this
work, we extend physical attacks to more challenging object detection models, a
broader class of deep learning algorithms widely used to detect and label
multiple objects within a scene. Improving upon a previous physical attack on
image classifiers, we create perturbed physical objects that are either ignored
or mislabeled by object detection models. We implement a Disappearance Attack,
in which we cause a Stop sign to "disappear" according to the detector-either
by covering thesign with an adversarial Stop sign poster, or by adding
adversarial stickers onto the sign. In a video recorded in a controlled lab
environment, the state-of-the-art YOLOv2 detector failed to recognize these
adversarial Stop signs in over 85% of the video frames. In an outdoor
experiment, YOLO was fooled by the poster and sticker attacks in 72.5% and
63.5% of the video frames respectively. We also use Faster R-CNN, a different
object detection model, to demonstrate the transferability of our adversarial
perturbations. The created poster perturbation is able to fool Faster R-CNN in
85.9% of the video frames in a controlled lab environment, and 40.2% of the
video frames in an outdoor environment. Finally, we present preliminary results
with a new Creation Attack, where in innocuous physical stickers fool a model
into detecting nonexistent objects.
|
Dialectical GAN for SAR Image Translation: From Sentinel-1 to TerraSAR-X
|
Contrary to optical images, Synthetic Aperture Radar (SAR) images are in
different electromagnetic spectrum where the human visual system is not
accustomed to. Thus, with more and more SAR applications, the demand for
enhanced high-quality SAR images has increased considerably. However,
high-quality SAR images entail high costs due to the limitations of current SAR
devices and their image processing resources. To improve the quality of SAR
images and to reduce the costs of their generation, we propose a Dialectical
Generative Adversarial Network (Dialectical GAN) to generate high-quality SAR
images. This method is based on the analysis of hierarchical SAR information
and the "dialectical" structure of GAN frameworks. As a demonstration, a
typical example will be shown where a low-resolution SAR image (e.g., a
Sentinel-1 image) with large ground coverage is translated into a
high-resolution SAR image (e.g., a TerraSAR-X image). Three traditional
algorithms are compared, and a new algorithm is proposed based on a network
framework by combining conditional WGAN-GP (Wasserstein Generative Adversarial
Network - Gradient Penalty) loss functions and Spatial Gram matrices under the
rule of dialectics. Experimental results show that the SAR image translation
works very well when we compare the results of our proposed method with the
selected traditional methods.
|
Escaping the Curse of Dimensionality in Similarity Learning: Efficient
Frank-Wolfe Algorithm and Generalization Bounds
|
Similarity and metric learning provides a principled approach to construct a
task-specific similarity from weakly supervised data. However, these methods
are subject to the curse of dimensionality: as the number of features grows
large, poor generalization is to be expected and training becomes intractable
due to high computational and memory costs. In this paper, we propose a
similarity learning method that can efficiently deal with high-dimensional
sparse data. This is achieved through a parameterization of similarity
functions by convex combinations of sparse rank-one matrices, together with the
use of a greedy approximate Frank-Wolfe algorithm which provides an efficient
way to control the number of active features. We show that the convergence rate
of the algorithm, as well as its time and memory complexity, are independent of
the data dimension. We further provide a theoretical justification of our
modeling choices through an analysis of the generalization error, which depends
logarithmically on the sparsity of the solution rather than on the number of
features. Our experiments on datasets with up to one million features
demonstrate the ability of our approach to generalize well despite the high
dimensionality as well as its superiority compared to several competing
methods.
|
Competition vs. Concatenation in Skip Connections of Fully Convolutional
Networks
|
Increased information sharing through short and long-range skip connections
between layers in fully convolutional networks have demonstrated significant
improvement in performance for semantic segmentation. In this paper, we propose
Competitive Dense Fully Convolutional Networks (CDFNet) by introducing
competitive maxout activations in place of naive feature concatenation for
inducing competition amongst layers. Within CDFNet, we propose two
architectural contributions, namely competitive dense block (CDB) and
competitive unpooling block (CUB) to induce competition at local and global
scales for short and long-range skip connections respectively. This extension
is demonstrated to boost learning of specialized sub-networks targeted at
segmenting specific anatomies, which in turn eases the training of complex
tasks. We present the proof-of-concept on the challenging task of whole body
segmentation in the publicly available VISCERAL benchmark and demonstrate
improved performance over multiple learning and registration based
state-of-the-art methods.
|
Wireless Multi-Sensor Networks for Smart Cities: A Prototype System with
Statistical Data Analysis
|
As urbanization proceeds at an astonishing rate, cities have to continuously
improve their solutions that affect the safety, health and overall wellbeing of
their residents. Smart city projects worldwide build on advanced sensor,
information and communication technologies to help dealing with issues like air
pollution, waste management, traffic optimization, and energy efficiency. The
paper reports about the prototype of a smart city initiative in Budapest which
applies various sensors installed on the public lighting system and a
cloud-based analytical module. While the installed wireless multi-sensor
network gathers information about a number of stressors, the module integrates
and statistically processes the data. The module can handle inconsistent,
missing and noisy data and can extrapolate the measurements in time and space,
namely, it can create short-term forecasts and smoothed maps, both accompanied
by reliability estimates. The resulting database uses geometric representations
and can serve as an information centre for public services.
|
TESSERACT: Eliminating Experimental Bias in Malware Classification
across Space and Time
|
Is Android malware classification a solved problem? Published F1 scores of up
to 0.99 appear to leave very little room for improvement. In this paper, we
argue that results are commonly inflated due to two pervasive sources of
experimental bias: "spatial bias" caused by distributions of training and
testing data that are not representative of a real-world deployment; and
"temporal bias" caused by incorrect time splits of training and testing sets,
leading to impossible configurations. We propose a set of space and time
constraints for experiment design that eliminates both sources of bias. We
introduce a new metric that summarizes the expected robustness of a classifier
in a real-world setting, and we present an algorithm to tune its performance.
Finally, we demonstrate how this allows us to evaluate mitigation strategies
for time decay such as active learning. We have implemented our solutions in
TESSERACT, an open source evaluation framework for comparing malware
classifiers in a realistic setting. We used TESSERACT to evaluate three Android
malware classifiers from the literature on a dataset of 129K applications
spanning over three years. Our evaluation confirms that earlier published
results are biased, while also revealing counter-intuitive performance and
showing that appropriate tuning can lead to significant improvements.
|
Talking Face Generation by Adversarially Disentangled Audio-Visual
Representation
|
Talking face generation aims to synthesize a sequence of face images that
correspond to a clip of speech. This is a challenging task because face
appearance variation and semantics of speech are coupled together in the subtle
movements of the talking face regions. Existing works either construct specific
face appearance model on specific subjects or model the transformation between
lip motion and speech. In this work, we integrate both aspects and enable
arbitrary-subject talking face generation by learning disentangled audio-visual
representation. We find that the talking face sequence is actually a
composition of both subject-related information and speech-related information.
These two spaces are then explicitly disentangled through a novel
associative-and-adversarial training process. This disentangled representation
has an advantage where both audio and video can serve as inputs for generation.
Extensive experiments show that the proposed approach generates realistic
talking face sequences on arbitrary subjects with much clearer lip motion
patterns than previous work. We also demonstrate the learned audio-visual
representation is extremely useful for the tasks of automatic lip reading and
audio-video retrieval.
|
The Deep Kernelized Autoencoder
|
Autoencoders learn data representations (codes) in such a way that the input
is reproduced at the output of the network. However, it is not always clear
what kind of properties of the input data need to be captured by the codes.
Kernel machines have experienced great success by operating via inner-products
in a theoretically well-defined reproducing kernel Hilbert space, hence
capturing topological properties of input data. In this paper, we enhance the
autoencoder's ability to learn effective data representations by aligning inner
products between codes with respect to a kernel matrix. By doing so, the
proposed kernelized autoencoder allows learning similarity-preserving
embeddings of input data, where the notion of similarity is explicitly
controlled by the user and encoded in a positive semi-definite kernel matrix.
Experiments are performed for evaluating both reconstruction and kernel
alignment performance in classification tasks and visualization of
high-dimensional data. Additionally, we show that our method is capable to
emulate kernel principal component analysis on a denoising task, obtaining
competitive results at a much lower computational cost.
|
Semi-Generative Modelling: Covariate-Shift Adaptation with Cause and
Effect Features
|
Current methods for covariate-shift adaptation use unlabelled data to compute
importance weights or domain-invariant features, while the final model is
trained on labelled data only. Here, we consider a particular case of covariate
shift which allows us also to learn from unlabelled data, that is, combining
adaptation with semi-supervised learning. Using ideas from causality, we argue
that this requires learning with both causes, $X_C$, and effects, $X_E$, of a
target variable, $Y$, and show how this setting leads to what we call a
semi-generative model, $P(Y,X_E|X_C,\theta)$. Our approach is robust to domain
shifts in the distribution of causal features and leverages unlabelled data by
learning a direct map from causes to effects. Experiments on synthetic data
demonstrate significant improvements in classification over purely-supervised
and importance-weighting baselines.
|
Boosting algorithms for uplift modeling
|
Uplift modeling is an area of machine learning which aims at predicting the
causal effect of some action on a given individual. The action may be a medical
procedure, marketing campaign, or any other circumstance controlled by the
experimenter. Building an uplift model requires two training sets: the
treatment group, where individuals have been subject to the action, and the
control group, where no action has been performed. An uplift model allows then
to assess the gain resulting from taking the action on a given individual, such
as the increase in probability of patient recovery or of a product being
purchased. This paper describes an adaptation of the well-known boosting
techniques to the uplift modeling case. We formulate three desirable properties
which an uplift boosting algorithm should have. Since all three properties
cannot be satisfied simultaneously, we propose three uplift boosting
algorithms, each satisfying two of them. Experiments demonstrate the usefulness
of the proposed methods, which often dramatically improve performance of the
base models and are thus new and powerful tools for uplift modeling.
|
Optimal Bounds on the VC-dimension
|
The VC-dimension of a set system is a way to capture its complexity and has
been a key parameter studied extensively in machine learning and geometry
communities. In this paper, we resolve two longstanding open problems on
bounding the VC-dimension of two fundamental set systems: $k$-fold
unions/intersections of half-spaces, and the simplices set system. Among other
implications, it settles an open question in machine learning that was first
studied in the 1989 foundational paper of Blumer, Ehrenfeucht, Haussler and
Warmuth as well as by Eisenstat and Angluin and Johnson.
|
Future Semantic Segmentation with Convolutional LSTM
|
We consider the problem of predicting semantic segmentation of future frames
in a video. Given several observed frames in a video, our goal is to predict
the semantic segmentation map of future frames that are not yet observed. A
reliable solution to this problem is useful in many applications that require
real-time decision making, such as autonomous driving. We propose a novel model
that uses convolutional LSTM (ConvLSTM) to encode the spatiotemporal
information of observed frames for future prediction. We also extend our model
to use bidirectional ConvLSTM to capture temporal information in both
directions. Our proposed approach outperforms other state-of-the-art methods on
the benchmark dataset.
|
A Fully Convolutional Neural Network Approach to End-to-End Speech
Enhancement
|
This paper will describe a novel approach to the cocktail party problem that
relies on a fully convolutional neural network (FCN) architecture. The FCN
takes noisy audio data as input and performs nonlinear, filtering operations to
produce clean audio data of the target speech at the output. Our method learns
a model for one specific speaker, and is then able to extract that speakers
voice from babble background noise. Results from experimentation indicate the
ability to generalize to new speakers and robustness to new noise environments
of varying signal-to-noise ratios. A potential application of this method would
be for use in hearing aids. A pre-trained model could be quickly fine tuned for
an individuals family members and close friends, and deployed onto a hearing
aid to assist listeners in noisy environments.
|
An Efficient End-to-End Neural Model for Handwritten Text Recognition
|
Offline handwritten text recognition from images is an important problem for
enterprises attempting to digitize large volumes of handmarked scanned
documents/reports. Deep recurrent models such as Multi-dimensional LSTMs have
been shown to yield superior performance over traditional Hidden Markov Model
based approaches that suffer from the Markov assumption and therefore lack the
representational power of RNNs. In this paper we introduce a novel approach
that combines a deep convolutional network with a recurrent Encoder-Decoder
network to map an image to a sequence of characters corresponding to the text
present in the image. The entire model is trained end-to-end using Focal Loss,
an improvement over the standard Cross-Entropy loss that addresses the class
imbalance problem, inherent to text recognition. To enhance the decoding
capacity of the model, Beam Search algorithm is employed which searches for the
best sequence out of a set of hypotheses based on a joint distribution of
individual characters. Our model takes as input a downsampled version of the
original image thereby making it both computationally and memory efficient. The
experimental results were benchmarked against two publicly available datasets,
IAM and RIMES. We surpass the state-of-the-art word level accuracy on the
evaluation set of both datasets by 3.5% & 1.1%, respectively.
|
Prior Convictions: Black-Box Adversarial Attacks with Bandits and Priors
|
We study the problem of generating adversarial examples in a black-box
setting in which only loss-oracle access to a model is available. We introduce
a framework that conceptually unifies much of the existing work on black-box
attacks, and we demonstrate that the current state-of-the-art methods are
optimal in a natural sense. Despite this optimality, we show how to improve
black-box attacks by bringing a new element into the problem: gradient priors.
We give a bandit optimization-based algorithm that allows us to seamlessly
integrate any such priors, and we explicitly identify and incorporate two
examples. The resulting methods use two to four times fewer queries and fail
two to five times less often than the current state-of-the-art.
|
Deep Learning
|
Deep learning (DL) is a high dimensional data reduction technique for
constructing high-dimensional predictors in input-output models. DL is a form
of machine learning that uses hierarchical layers of latent features. In this
article, we review the state-of-the-art of deep learning from a modeling and
algorithmic perspective. We provide a list of successful areas of applications
in Artificial Intelligence (AI), Image Processing, Robotics and Automation.
Deep learning is predictive in its nature rather then inferential and can be
viewed as a black-box methodology for high-dimensional function estimation.
|
Convolutional Neural Networks Analyzed via Inverse Problem Theory and
Sparse Representations
|
Inverse problems in imaging such as denoising, deblurring, superresolution
(SR) have been addressed for many decades. In recent years, convolutional
neural networks (CNNs) have been widely used for many inverse problem areas.
Although their indisputable success, CNNs are not mathematically validated as
to how and what they learn. In this paper, we prove that during training, CNN
elements solve for inverse problems which are optimum solutions stored as CNN
neuron filters. We discuss the necessity of mutual coherence between CNN layer
elements in order for a network to converge to the optimum solution. We prove
that required mutual coherence can be provided by the usage of residual
learning and skip connections. We have set rules over training sets and depth
of networks for better convergence, i.e. performance.
|
signProx: One-Bit Proximal Algorithm for Nonconvex Stochastic
Optimization
|
Stochastic gradient descent (SGD) is one of the most widely used optimization
methods for parallel and distributed processing of large datasets. One of the
key limitations of distributed SGD is the need to regularly communicate the
gradients between different computation nodes. To reduce this communication
bottleneck, recent work has considered a one-bit variant of SGD, where only the
sign of each gradient element is used in optimization. In this paper, we extend
this idea by proposing a stochastic variant of the proximal-gradient method
that also uses one-bit per update element. We prove the theoretical convergence
of the method for non-convex optimization under a set of explicit assumptions.
Our results indicate that the compressed method can match the convergence rate
of the uncompressed one, making the proposed method potentially appealing for
distributed processing of large datasets.
|
A Fast, Principled Working Set Algorithm for Exploiting Piecewise Linear
Structure in Convex Problems
|
By reducing optimization to a sequence of smaller subproblems, working set
algorithms achieve fast convergence times for many machine learning problems.
Despite such performance, working set implementations often resort to
heuristics to determine subproblem size, makeup, and stopping criteria. We
propose BlitzWS, a working set algorithm with useful theoretical guarantees.
Our theory relates subproblem size and stopping criteria to the amount of
progress during each iteration. This result motivates strategies for optimizing
algorithmic parameters and discarding irrelevant components as BlitzWS
progresses toward a solution. BlitzWS applies to many convex problems,
including training L1-regularized models and support vector machines. We
showcase this versatility with empirical comparisons, which demonstrate BlitzWS
is indeed a fast algorithm.
|
Baidu Apollo EM Motion Planner
|
In this manuscript, we introduce a real-time motion planning system based on
the Baidu Apollo (open source) autonomous driving platform. The developed
system aims to address the industrial level-4 motion planning problem while
considering safety, comfort and scalability. The system covers multilane and
single-lane autonomous driving in a hierarchical manner: (1) The top layer of
the system is a multilane strategy that handles lane-change scenarios by
comparing lane-level trajectories computed in parallel. (2) Inside the
lane-level trajectory generator, it iteratively solves path and speed
optimization based on a Frenet frame. (3) For path and speed optimization, a
combination of dynamic programming and spline-based quadratic programming is
proposed to construct a scalable and easy-to-tune framework to handle traffic
rules, obstacle decisions and smoothness simultaneously. The planner is
scalable to both highway and lower-speed city driving scenarios. We also
demonstrate the algorithm through scenario illustrations and on-road test
results.
The system described in this manuscript has been deployed to dozens of Baidu
Apollo autonomous driving vehicles since Apollo v1.5 was announced in September
2017. As of May 16th, 2018, the system has been tested under 3,380 hours and
approximately 68,000 kilometers (42,253 miles) of closed-loop autonomous
driving under various urban scenarios.
The algorithm described in this manuscript is available at
https://github.com/ApolloAuto/apollo/tree/master/modules/planning.
|
Learning Heuristics for Quantified Boolean Formulas through Deep
Reinforcement Learning
|
We demonstrate how to learn efficient heuristics for automated reasoning
algorithms for quantified Boolean formulas through deep reinforcement learning.
We focus on a backtracking search algorithm, which can already solve formulas
of impressive size - up to hundreds of thousands of variables. The main
challenge is to find a representation of these formulas that lends itself to
making predictions in a scalable way. For a family of challenging problems, we
learned a heuristic that solves significantly more formulas compared to the
existing handwritten heuristics.
|
Learning Optimal Resource Allocations in Wireless Systems
|
This paper considers the design of optimal resource allocation policies in
wireless communication systems which are generically modeled as a functional
optimization problem with stochastic constraints. These optimization problems
have the structure of a learning problem in which the statistical loss appears
as a constraint, motivating the development of learning methodologies to
attempt their solution. To handle stochastic constraints, training is
undertaken in the dual domain. It is shown that this can be done with small
loss of optimality when using near-universal learning parameterizations. In
particular, since deep neural networks (DNN) are near-universal their use is
advocated and explored. DNNs are trained here with a model-free primal-dual
method that simultaneously learns a DNN parametrization of the resource
allocation policy and optimizes the primal and dual variables. Numerical
simulations demonstrate the strong performance of the proposed approach on a
number of common wireless resource allocation problems.
|
Streaming Methods for Restricted Strongly Convex Functions with
Applications to Prototype Selection
|
In this paper, we show that if the optimization function is
restricted-strongly-convex (RSC) and restricted-smooth (RSM) -- a rich subclass
of weakly submodular functions -- then a streaming algorithm with constant
factor approximation guarantee is possible. More generally, our results are
applicable to any monotone weakly submodular function with submodularity ratio
bounded from above. This (positive) result which provides a sufficient
condition for having a constant factor streaming guarantee for weakly
submodular functions may be of special interest given the recent negative
result (Elenberg et al., 2017) for the general class of weakly submodular
functions. We apply our streaming algorithms for creating compact synopsis of
large complex datasets, by selecting $m$ representative elements, by optimizing
a suitable RSC and RSM objective function. Above results hold even with
additional constraints such as learning non-negative weights, for
interpretability, for each selected element indicative of its importance. We
empirically evaluate our algorithms on two real datasets: MNIST- a handwritten
digits dataset and Letters- a UCI dataset containing the alphabet written in
different fonts and styles. We observe that our algorithms are orders of
magnitude faster than the state-of-the-art streaming algorithm for weakly
submodular functions and with our main algorithm still providing equally good
solutions in practice.
|
Simultaneous Adversarial Training - Learn from Others Mistakes
|
Adversarial examples are maliciously tweaked images that can easily fool
machine learning techniques, such as neural networks, but they are normally not
visually distinguishable for human beings. One of the main approaches to solve
this problem is to retrain the networks using those adversarial examples,
namely adversarial training. However, standard adversarial training might not
actually change the decision boundaries but cause the problem of gradient
masking, resulting in a weaker ability to generate adversarial examples.
Therefore, it cannot alleviate the problem of black-box attacks, where
adversarial examples generated from other networks can transfer to the targeted
one. In order to reduce the problem of black-box attacks, we propose a novel
method that allows two networks to learn from each others' adversarial examples
and become resilient to black-box attacks. We also combine this method with a
simple domain adaptation to further improve the performance.
|
What is not where: the challenge of integrating spatial representations
into deep learning architectures
|
This paper examines to what degree current deep learning architectures for
image caption generation capture spatial language. On the basis of the
evaluation of examples of generated captions from the literature we argue that
systems capture what objects are in the image data but not where these objects
are located: the captions generated by these systems are the output of a
language model conditioned on the output of an object detector that cannot
capture fine-grained location information. Although language models provide
useful knowledge for image captions, we argue that deep learning image
captioning architectures should also model geometric relations between objects.
|
On the Analysis of Trajectories of Gradient Descent in the Optimization
of Deep Neural Networks
|
Theoretical analysis of the error landscape of deep neural networks has
garnered significant interest in recent years. In this work, we theoretically
study the importance of noise in the trajectories of gradient descent towards
optimal solutions in multi-layer neural networks. We show that adding noise (in
different ways) to a neural network while training increases the rank of the
product of weight matrices of a multi-layer linear neural network. We thus
study how adding noise can assist reaching a global optimum when the product
matrix is full-rank (under certain conditions). We establish theoretical
foundations between the noise induced into the neural network - either to the
gradient, to the architecture, or to the input/output to a neural network - and
the rank of product of weight matrices. We corroborate our theoretical findings
with empirical results.
|
Linear density-based clustering with a discrete density model
|
Density-based clustering techniques are used in a wide range of data mining
applications. One of their most attractive features con- sists in not making
use of prior knowledge of the number of clusters that a dataset contains along
with their shape. In this paper we propose a new algorithm named Linear DBSCAN
(Lin-DBSCAN), a simple approach to clustering inspired by the density model
introduced with the well known algorithm DBSCAN. Designed to minimize the
computational cost of density based clustering on geospatial data, Lin-DBSCAN
features a linear time complexity that makes it suitable for real-time
applications on low-resource devices. Lin-DBSCAN uses a discrete version of the
density model of DBSCAN that takes ad- vantage of a grid-based scan and merge
approach. The name of the algorithm stems exactly from its main features
outlined above. The algorithm was tested with well known data sets.
Experimental results prove the efficiency and the validity of this approach
over DBSCAN in the context of spatial data clustering, enabling the use of a
density-based clustering technique on large datasets with low computational
cost.
|
Recent Advances in Deep Learning: An Overview
|
Deep Learning is one of the newest trends in Machine Learning and Artificial
Intelligence research. It is also one of the most popular scientific research
trends now-a-days. Deep learning methods have brought revolutionary advances in
computer vision and machine learning. Every now and then, new and new deep
learning techniques are being born, outperforming state-of-the-art machine
learning and even existing deep learning techniques. In recent years, the world
has seen many major breakthroughs in this field. Since deep learning is
evolving at a huge speed, its kind of hard to keep track of the regular
advances especially for new researchers. In this paper, we are going to briefly
discuss about recent advances in Deep Learning for past few years.
|
Predicting purchasing intent: Automatic Feature Learning using Recurrent
Neural Networks
|
We present a neural network for predicting purchasing intent in an Ecommerce
setting. Our main contribution is to address the significant investment in
feature engineering that is usually associated with state-of-the-art methods
such as Gradient Boosted Machines. We use trainable vector spaces to model
varied, semi-structured input data comprising categoricals, quantities and
unique instances. Multi-layer recurrent neural networks capture both
session-local and dataset-global event dependencies and relationships for user
sessions of any length. An exploration of model design decisions including
parameter sharing and skip connections further increase model accuracy. Results
on benchmark datasets deliver classification accuracy within 98% of
state-of-the-art on one and exceed state-of-the-art on the second without the
need for any domain / dataset-specific feature engineering on both short and
long event sequences.
|
Sign-Perturbed Sums: A New System Identification Approach for
Constructing Exact Non-Asymptotic Confidence Regions in Linear Regression
Models
|
We propose a new system identification method, called Sign-Perturbed Sums
(SPS), for constructing non-asymptotic confidence regions under mild
statistical assumptions. SPS is introduced for linear regression models,
including but not limited to FIR systems, and we show that the SPS confidence
regions have exact confidence probabilities, i.e., they contain the true
parameter with a user-chosen exact probability for any finite data set.
Moreover, we also prove that the SPS regions are star convex with the
Least-Squares (LS) estimate as a star center. The main assumptions of SPS are
that the noise terms are independent and symmetrically distributed about zero,
but they can be nonstationary, and their distributions need not be known. The
paper also proposes a computationally efficient ellipsoidal outer approximation
algorithm for SPS. Finally, SPS is demonstrated through a number of simulation
experiments.
|
A Preliminary Study On the Sustainability of Android Malware Detection
|
Machine learning-based malware detection dominates current security defense
approaches for Android apps. However, due to the evolution of Android platforms
and malware, existing such techniques are widely limited by their need for
constant retraining that are costly, and reliance on new malware samples that
may not be timely available. As a result, new and emerging malware slips
through, as seen from the continued surging of malware in the wild. Thus, a
more practical detector needs not only to be accurate but, more critically, to
be able to sustain its capabilities over time without frequent retraining. In
this paper, we study how Android apps evolve as a population over time, in
terms of their behaviors related to accesses to sensitive information and
operations. We first perform a longitudinal characterization of 6K benign and
malicious apps developed across seven years, with focus on these sensitive
accesses in app executions. Our study reveals, during the long evolution, a
consistent, clear differentiation between malware and benign apps regarding
such accesses, measured by relative statistics of relevant method calls.
Following these findings, we developed DroidSpan, a novel classification system
based on a new behavioral profile for Android apps. Through an extensive
evaluation, we showed that DroidSpan can not only effectively detect malware
but sustain high detection accuracy (93% F1 measure) for four years (with 81%
F1 for five years). Through a dedicated study, we also showed its resiliency to
sophisticated evasion schemes. By comparing to a state-of-the-art malware
detector, we demonstrated the largely superior sustainability of our approach
at reasonable costs.
|
Rapid Autonomous Car Control based on Spatial and Temporal Visual Cues
|
We present a novel approach to modern car control utilizing a combination of
Deep Convolutional Neural Networks and Long Short-Term Memory Systems: Both of
which are a subsection of Hierarchical Representations Learning, more commonly
known as Deep Learning. Using Deep Convolutional Neural Networks and Long
Short-Term Memory Systems (DCNN/LSTM), we propose an end-to-end approach to
accurately predict steering angles and throttle values. We use this algorithm
on our latest robot, El Toro Grande 1 (ETG) which is equipped with a variety of
sensors in order to localize itself in its environment. Using previous training
data and the data that it collects during circuit and drag races, it predicts
throttle and steering angles in order to stay on path and avoid colliding into
other robots. This allows ETG to theoretically race on any track with
sufficient training data.
|
Learning Deep Hidden Nonlinear Dynamics from Aggregate Data
|
Learning nonlinear dynamics from diffusion data is a challenging problem
since the individuals observed may be different at different time points,
generally following an aggregate behaviour. Existing work cannot handle the
tasks well since they model such dynamics either directly on observations or
enforce the availability of complete longitudinal individual-level
trajectories. However, in most of the practical applications, these
requirements are unrealistic: the evolving dynamics may be too complex to be
modeled directly on observations, and individual-level trajectories may not be
available due to technical limitations, experimental costs and/or privacy
issues. To address these challenges, we formulate a model of diffusion dynamics
as the {\em hidden stochastic process} via the introduction of hidden variables
for flexibility, and learn the hidden dynamics directly on {\em aggregate
observations} without any requirement for individual-level trajectories. We
propose a dynamic generative model with Wasserstein distance for LEarninG dEep
hidden Nonlinear Dynamics (LEGEND) and prove its theoretical guarantees as
well. Experiments on a range of synthetic and real-world datasets illustrate
that LEGEND has very strong performance compared to state-of-the-art baselines.
|
NAVREN-RL: Learning to fly in real environment via end-to-end deep
reinforcement learning using monocular images
|
We present NAVREN-RL, an approach to NAVigate an unmanned aerial vehicle in
an indoor Real ENvironment via end-to-end reinforcement learning RL. A suitable
reward function is designed keeping in mind the cost and weight constraints for
micro drone with minimum number of sensing modalities. Collection of small
number of expert data and knowledge based data aggregation is integrated into
the RL process to aid convergence. Experimentation is carried out on a Parrot
AR drone in different indoor arenas and the results are compared with other
baseline technologies. We demonstrate how the drone successfully avoids
obstacles and navigates across different arenas.
|
Deep learning at the shallow end: Malware classification for non-domain
experts
|
Current malware detection and classification approaches generally rely on
time consuming and knowledge intensive processes to extract patterns
(signatures) and behaviors from malware, which are then used for
identification. Moreover, these signatures are often limited to local,
contiguous sequences within the data whilst ignoring their context in relation
to each other and throughout the malware file as a whole. We present a Deep
Learning based malware classification approach that requires no expert domain
knowledge and is based on a purely data driven approach for complex pattern and
feature identification.
|
Multi-scale Alignment and Contextual History for Attention Mechanism in
Sequence-to-sequence Model
|
A sequence-to-sequence model is a neural network module for mapping two
sequences of different lengths. The sequence-to-sequence model has three core
modules: encoder, decoder, and attention. Attention is the bridge that connects
the encoder and decoder modules and improves model performance in many tasks.
In this paper, we propose two ideas to improve sequence-to-sequence model
performance by enhancing the attention module. First, we maintain the history
of the location and the expected context from several previous time-steps.
Second, we apply multiscale convolution from several previous attention vectors
to the current decoder state. We utilized our proposed framework for
sequence-to-sequence speech recognition and text-to-speech systems. The results
reveal that our proposed extension could improve performance significantly
compared to a standard attention baseline.
|
Unified Hypersphere Embedding for Speaker Recognition
|
Incremental improvements in accuracy of Convolutional Neural Networks are
usually achieved through use of deeper and more complex models trained on
larger datasets. However, enlarging dataset and models increases the
computation and storage costs and cannot be done indefinitely. In this work, we
seek to improve the identification and verification accuracy of a
text-independent speaker recognition system without use of extra data or deeper
and more complex models by augmenting the training and testing data, finding
the optimal dimensionality of embedding space and use of more discriminative
loss functions. Results of experiments on VoxCeleb dataset suggest that: (i)
Simple repetition and random time-reversion of utterances can reduce prediction
errors by up to 18%. (ii) Lower dimensional embeddings are more suitable for
verification. (iii) Use of proposed logistic margin loss function leads to
unified embeddings with state-of-the-art identification and competitive
verification accuracies.
|
Accelerated Structure-Aware Reinforcement Learning for Delay-Sensitive
Energy Harvesting Wireless Sensors
|
We investigate an energy-harvesting wireless sensor transmitting
latency-sensitive data over a fading channel. The sensor injects captured data
packets into its transmission queue and relies on ambient energy harvested from
the environment to transmit them. We aim to find the optimal scheduling policy
that decides whether or not to transmit the queue's head-of-line packet at each
transmission opportunity such that the expected packet queuing delay is
minimized given the available harvested energy. No prior knowledge of the
stochastic processes that govern the channel, captured data, or harvested
energy dynamics are assumed, thereby necessitating the use of online learning
to optimize the scheduling policy. We formulate this scheduling problem as a
Markov decision process (MDP) and analyze the structural properties of its
optimal value function. In particular, we show that it is non-decreasing and
has increasing differences in the queue backlog and that it is non-increasing
and has increasing differences in the battery state. We exploit this structure
to formulate a novel accelerated reinforcement learning (RL) algorithm to solve
the scheduling problem online at a much faster learning rate, while limiting
the induced computational complexity. Our experiments demonstrate that the
proposed algorithm closely approximates the performance of an optimal offline
solution that requires a priori knowledge of the channel, captured data, and
harvested energy dynamics. Simultaneously, by leveraging the value function's
structure, our approach achieves competitive performance relative to a
state-of-the-art RL algorithm, at potentially orders of magnitude lower
complexity. Finally, considerable performance gains are demonstrated over the
well-known and widely used Q-learning algorithm.
|
MOBA-Slice: A Time Slice Based Evaluation Framework of Relative
Advantage between Teams in MOBA Games
|
Multiplayer Online Battle Arena (MOBA) is currently one of the most popular
genres of digital games around the world. The domain of knowledge contained in
these complicated games is large. It is hard for humans and algorithms to
evaluate the real-time game situation or predict the game result. In this
paper, we introduce MOBA-Slice, a time slice based evaluation framework of
relative advantage between teams in MOBA games. MOBA-Slice is a quantitative
evaluation method based on learning, similar to the value network of AlphaGo.
It establishes a foundation for further MOBA related research including AI
development. In MOBA-Slice, with an analysis of the deciding factors of MOBA
game results, we design a neural network model to fit our discounted evaluation
function. Then we apply MOBA-Slice to Defense of the Ancients 2 (DotA2), a
typical and popular MOBA game. Experiments on a large number of match replays
show that our model works well on arbitrary matches. MOBA-Slice not only has an
accuracy 3.7% higher than DotA Plus Assistant at result prediction, but also
supports the prediction of the remaining time of the game, and then realizes
the evaluation of relative advantage between teams.
|
An Intersectional Definition of Fairness
|
We propose definitions of fairness in machine learning and artificial
intelligence systems that are informed by the framework of intersectionality, a
critical lens arising from the Humanities literature which analyzes how
interlocking systems of power and oppression affect individuals along
overlapping dimensions including gender, race, sexual orientation, class, and
disability. We show that our criteria behave sensibly for any subset of the set
of protected attributes, and we prove economic, privacy, and generalization
guarantees. We provide a learning algorithm which respects our intersectional
fairness criteria. Case studies on census data and the COMPAS criminal
recidivism dataset demonstrate the utility of our methods.
|
EnsembleDAgger: A Bayesian Approach to Safe Imitation Learning
|
While imitation learning is often used in robotics, the approach frequently
suffers from data mismatch and compounding errors. DAgger is an iterative
algorithm that addresses these issues by aggregating training data from both
the expert and novice policies, but does not consider the impact of safety. We
present a probabilistic extension to DAgger, which attempts to quantify the
confidence of the novice policy as a proxy for safety. Our method,
EnsembleDAgger, approximates a Gaussian Process using an ensemble of neural
networks. Using the variance as a measure of confidence, we compute a decision
rule that captures how much we doubt the novice, thus determining when it is
safe to allow the novice to act. With this approach, we aim to maximize the
novice's share of actions, while constraining the probability of failure. We
demonstrate improved safety and learning performance compared to other DAgger
variants and classic imitation learning on an inverted pendulum and in the
MuJoCo HalfCheetah environment.
|
Knowledge-based Transfer Learning Explanation
|
Machine learning explanation can significantly boost machine learning's
application in decision making, but the usability of current methods is limited
in human-centric explanation, especially for transfer learning, an important
machine learning branch that aims at utilizing knowledge from one learning
domain (i.e., a pair of dataset and prediction task) to enhance prediction
model training in another learning domain. In this paper, we propose an
ontology-based approach for human-centric explanation of transfer learning.
Three kinds of knowledge-based explanatory evidence, with different
granularities, including general factors, particular narrators and core
contexts are first proposed and then inferred with both local ontologies and
external knowledge bases. The evaluation with US flight data and DBpedia has
presented their confidence and availability in explaining the transferability
of feature representation in flight departure delay forecasting.
|
PaloBoost: An Overfitting-robust TreeBoost with Out-of-Bag Sample
Regularization Techniques
|
Stochastic Gradient TreeBoost is often found in many winning solutions in
public data science challenges. Unfortunately, the best performance requires
extensive parameter tuning and can be prone to overfitting. We propose
PaloBoost, a Stochastic Gradient TreeBoost model that uses novel regularization
techniques to guard against overfitting and is robust to parameter settings.
PaloBoost uses the under-utilized out-of-bag samples to perform gradient-aware
pruning and estimate adaptive learning rates. Unlike other Stochastic Gradient
TreeBoost models that use the out-of-bag samples to estimate test errors,
PaloBoost treats the samples as a second batch of training samples to prune the
trees and adjust the learning rates. As a result, PaloBoost can dynamically
adjust tree depths and learning rates to achieve faster learning at the start
and slower learning as the algorithm converges. We illustrate how these
regularization techniques can be efficiently implemented and propose a new
formula for calculating feature importance to reflect the node coverages and
learning rates. Extensive experimental results on seven datasets demonstrate
that PaloBoost is robust to overfitting, is less sensitivity to the parameters,
and can also effectively identify meaningful features.
|
Real-Time 2D-3D Deformable Registration with Deep Learning and
Application to Lung Radiotherapy Targeting
|
Radiation therapy presents a need for dynamic tracking of a target tumor
volume. Fiducial markers such as implanted gold seeds have been used to gate
radiation delivery but the markers are invasive and gating significantly
increases treatment time. Pretreatment acquisition of a respiratory correlated
4DCT allows for determination of accurate motion tracking which is useful in
treatment planning. We design a patient-specific motion subspace and a deep
convolutional neural network to recover anatomical positions from a single
fluoroscopic projection in real-time. We use this deep network to approximate
the nonlinear inverse of a diffeomorphic deformation composed with radiographic
projection. This network recovers subspace coordinates to define the
patient-specific deformation of the lungs from a baseline anatomic position.
The geometric accuracy of the subspace deformations on real patient data is
similar to accuracy attained by original image registration between individual
respiratory-phase image volumes.
|
Score Permutation Based Finite Sample Inference for Generalized
AutoRegressive Conditional Heteroskedasticity (GARCH) Models
|
A standard model of (conditional) heteroscedasticity, i.e., the phenomenon
that the variance of a process changes over time, is the Generalized
AutoRegressive Conditional Heteroskedasticity (GARCH) model, which is
especially important for economics and finance. GARCH models are typically
estimated by the Quasi-Maximum Likelihood (QML) method, which works under mild
statistical assumptions. Here, we suggest a finite sample approach, called
ScoPe, to construct distribution-free confidence regions around the QML
estimate, which have exact coverage probabilities, despite no additional
assumptions about moments are made. ScoPe is inspired by the recently developed
Sign-Perturbed Sums (SPS) method, which however cannot be applied in the GARCH
case. ScoPe works by perturbing the score function using randomly permuted
residuals. This produces alternative samples which lead to exact confidence
regions. Experiments on simulated and stock market data are also presented, and
ScoPe is compared with the asymptotic theory and bootstrap approaches.
|
Visual Mesh: Real-time Object Detection Using Constant Sample Density
|
This paper proposes an enhancement of convolutional neural networks for
object detection in resource-constrained robotics through a geometric input
transformation called Visual Mesh. It uses object geometry to create a graph in
vision space, reducing computational complexity by normalizing the pixel and
feature density of objects. The experiments compare the Visual Mesh with
several other fast convolutional neural networks. The results demonstrate
execution times sixteen times quicker than the fastest competitor tested, while
achieving outstanding accuracy.
|
Aligning Points to Lines: Provable Approximations
|
We suggest a new optimization technique for minimizing the sum $\sum_{i=1}^n
f_i(x)$ of $n$ non-convex real functions that satisfy a property that we call
piecewise log-Lipschitz. This is by forging links between techniques in
computational geometry, combinatorics and convex optimization. As an example
application, we provide the first constant-factor approximation algorithms
whose running-time is polynomial in $n$ for the fundamental problem of
\emph{Points-to-Lines alignment}: Given $n$ points $p_1,\cdots,p_n$ and $n$
lines $\ell_1,\cdots,\ell_n$ on the plane and $z>0$, compute the matching
$\pi:[n]\to[n]$ and alignment (rotation matrix $R$ and a translation vector
$t$) that minimize the sum of Euclidean distances $\sum_{i=1}^n
\mathrm{dist}(Rp_i-t,\ell_{\pi(i)})^z$ between each point to its corresponding
line.
This problem is non-trivial even if $z=1$ and the matching $\pi$ is given. If
$\pi$ is given, the running time of our algorithms is $O(n^3)$, and even
near-linear in $n$ using core-sets that support: streaming, dynamic, and
distributed parallel computations in poly-logarithmic update time.
Generalizations for handling e.g. outliers or pseudo-distances such as
$M$-estimators for the problem are also provided.
Experimental results and open source code show that our provable algorithms
improve existing heuristics also in practice. A companion demonstration video
in the context of Augmented Reality shows how such algorithms may be used in
real-time systems.
|
LinkNBed: Multi-Graph Representation Learning with Entity Linkage
|
Knowledge graphs have emerged as an important model for studying complex
multi-relational data. This has given rise to the construction of numerous
large scale but incomplete knowledge graphs encoding information extracted from
various resources. An effective and scalable approach to jointly learn over
multiple graphs and eventually construct a unified graph is a crucial next step
for the success of knowledge-based inference for many downstream applications.
To this end, we propose LinkNBed, a deep relational learning framework that
learns entity and relationship representations across multiple graphs. We
identify entity linkage across graphs as a vital component to achieve our goal.
We design a novel objective that leverage entity linkage and build an efficient
multi-task training procedure. Experiments on link prediction and entity
linkage demonstrate substantial improvements over the state-of-the-art
relational learning approaches.
|
Learning to Play Pong using Policy Gradient Learning
|
Activities in reinforcement learning (RL) revolve around learning the Markov
decision process (MDP) model, in particular, the following parameters: state
values, V; state-action values, Q; and policy, pi. These parameters are
commonly implemented as an array. Scaling up the problem means scaling up the
size of the array and this will quickly lead to a computational bottleneck. To
get around this, the RL problem is commonly formulated to learn a specific task
using hand-crafted input features to curb the size of the array. In this
report, we discuss an alternative end-to-end Deep Reinforcement Learning (DRL)
approach where the DRL attempts to learn general task representations which in
our context refers to learning to play the Pong game from a sequence of screen
snapshots without game-specific hand-crafted features. We apply artificial
neural networks (ANN) to approximate a policy of the RL model. The policy
network, via Policy Gradients (PG) method, learns to play the Pong game from a
sequence of frames without any extra semantics apart from the pixel information
and the score. In contrast to the traditional tabular RL approach where the
contents in the array have clear interpretations such as V or Q, the
interpretation of knowledge content from the weights of the policy network is
more illusive. In this work, we experiment with various Deep ANN architectures
i.e., Feed forward ANN (FFNN), Convolution ANN (CNN) and Asynchronous Advantage
Actor-Critic (A3C). We also examine the activation of hidden nodes and the
weights between the input and the hidden layers, before and after the DRL has
successfully learnt to play the Pong game. Insights into the internal learning
mechanisms and future research directions are then discussed.
|
2P-DNN : Privacy-Preserving Deep Neural Networks Based on Homomorphic
Cryptosystem
|
Machine Learning as a Service (MLaaS), such as Microsoft Azure, Amazon AWS,
offers an effective DNN model to complete the machine learning task for small
businesses and individuals who are restricted to the lacking data and computing
power. However, here comes an issue that user privacy is ex-posed to the MLaaS
server, since users need to upload their sensitive data to the MLaaS server. In
order to preserve their privacy, users can encrypt their data before uploading
it. This makes it difficult to run the DNN model because it is not designed for
running in ciphertext domain. In this paper, using the Paillier homomorphic
cryptosystem we present a new Privacy-Preserving Deep Neural Network model that
we called 2P-DNN. This model can fulfill the machine leaning task in ciphertext
domain. By using 2P-DNN, MLaaS is able to provide a Privacy-Preserving machine
learning ser-vice for users. We build our 2P-DNN model based on LeNet-5, and
test it with the encrypted MNIST dataset. The classification accuracy is more
than 97%, which is close to the accuracy of LeNet-5 running with the MNIST
dataset and higher than that of other existing Privacy-Preserving machine
learning models
|
Multimodal Social Media Analysis for Gang Violence Prevention
|
Gang violence is a severe issue in major cities across the U.S. and recent
studies [Patton et al. 2017] have found evidence of social media communications
that can be linked to such violence in communities with high rates of exposure
to gang activity. In this paper we partnered computer scientists with social
work researchers, who have domain expertise in gang violence, to analyze how
public tweets with images posted by youth who mention gang associations on
Twitter can be leveraged to automatically detect psychosocial factors and
conditions that could potentially assist social workers and violence outreach
workers in prevention and early intervention programs. To this end, we
developed a rigorous methodology for collecting and annotating tweets. We
gathered 1,851 tweets and accompanying annotations related to visual concepts
and the psychosocial codes: aggression, loss, and substance use. These codes
are relevant to social work interventions, as they represent possible pathways
to violence on social media. We compare various methods for classifying tweets
into these three classes, using only the text of the tweet, only the image of
the tweet, or both modalities as input to the classifier. In particular, we
analyze the usefulness of mid-level visual concepts and the role of different
modalities for this tweet classification task. Our experiments show that
individually, text information dominates classification performance of the loss
class, while image information dominates the aggression and substance use
classes. Our multimodal approach provides a very promising improvement (18%
relative in mean average precision) over the best single modality approach.
Finally, we also illustrate the complexity of understanding social media data
and elaborate on open challenges.
|
Domain Generalization via Conditional Invariant Representation
|
Domain generalization aims to apply knowledge gained from multiple labeled
source domains to unseen target domains. The main difficulty comes from the
dataset bias: training data and test data have different distributions, and the
training set contains heterogeneous samples from different distributions. Let
$X$ denote the features, and $Y$ be the class labels. Existing domain
generalization methods address the dataset bias problem by learning a
domain-invariant representation $h(X)$ that has the same marginal distribution
$\mathbb{P}(h(X))$ across multiple source domains. The functional relationship
encoded in $\mathbb{P}(Y|X)$ is usually assumed to be stable across domains
such that $\mathbb{P}(Y|h(X))$ is also invariant. However, it is unclear
whether this assumption holds in practical problems. In this paper, we consider
the general situation where both $\mathbb{P}(X)$ and $\mathbb{P}(Y|X)$ can
change across all domains. We propose to learn a feature representation which
has domain-invariant class conditional distributions $\mathbb{P}(h(X)|Y)$. With
the conditional invariant representation, the invariance of the joint
distribution $\mathbb{P}(h(X),Y)$ can be guaranteed if the class prior
$\mathbb{P}(Y)$ does not change across training and test domains. Extensive
experiments on both synthetic and real data demonstrate the effectiveness of
the proposed method.
|
Risk Bounds for Unsupervised Cross-Domain Mapping with IPMs
|
The recent empirical success of unsupervised cross-domain mapping algorithms,
between two domains that share common characteristics, is not well-supported by
theoretical justifications. This lacuna is especially troubling, given the
clear ambiguity in such mappings.
We work with adversarial training methods based on IPMs and derive a novel
risk bound, which upper bounds the risk between the learned mapping $h$ and the
target mapping $y$, by a sum of three terms: (i) the risk between $h$ and the
most distant alternative mapping that was learned by the same cross-domain
mapping algorithm, (ii) the minimal discrepancy between the target domain and
the domain obtained by applying a hypothesis $h^*$ on the samples of the source
domain, where $h^*$ is a hypothesis selectable by the same algorithm. The bound
is directly related to Occam's razor and encourages the selection of the
minimal architecture that supports a small mapping discrepancy and (iii) an
approximation error term that decreases as the complexity of the class of
discriminators increases and is empirically shown to be small.
The bound leads to multiple algorithmic consequences, including a method for
hyperparameters selection and for early stopping in cross-domain mapping GANs.
We also demonstrate a novel capability for unsupervised learning of estimating
confidence in the mapping of every specific sample.
|
Implementing Neural Turing Machines
|
Neural Turing Machines (NTMs) are an instance of Memory Augmented Neural
Networks, a new class of recurrent neural networks which decouple computation
from memory by introducing an external memory unit. NTMs have demonstrated
superior performance over Long Short-Term Memory Cells in several sequence
learning tasks. A number of open source implementations of NTMs exist but are
unstable during training and/or fail to replicate the reported performance of
NTMs. This paper presents the details of our successful implementation of a
NTM. Our implementation learns to solve three sequential learning tasks from
the original NTM paper. We find that the choice of memory contents
initialization scheme is crucial in successfully implementing a NTM. Networks
with memory contents initialized to small constant values converge on average 2
times faster than the next best memory contents initialization scheme.
|
Particle Filtering Methods for Stochastic Optimization with Application
to Large-Scale Empirical Risk Minimization
|
This paper is concerned with sequential filtering based stochastic
optimization (FSO) approaches that leverage a probabilistic perspective to
implement the incremental proximity method (IPM). The present FSO methods are
derived based on the Kalman filter (KF) and the extended KF (EKF). In contrast
with typical methods such as stochastic gradient descent (SGD) and IPMs, they
do not need to pre-schedule the learning rate for convergence. Nevertheless,
they have limitations that inherit from the KF mechanism. As the particle
filtering (PF) method outperforms KF and its variants remarkably for nonlinear
non-Gaussian sequential filtering problems, it is natural to ask if FSO methods
can benefit from PF to get around of their limitations. We provide an
affirmative answer to this question by developing two PF based stochastic
optimizers (PFSOs). For performance evaluation, we apply them to address
nonlinear least-square fitting with simulated data, and empirical risk
minimization for binary classification of real data sets. Experimental results
demonstrate that PFSOs outperform remarkably a benchmark SGD algorithm, the
vanilla IPM, and KF-type FSO methods in terms of numerical stability,
convergence speed, and flexibility in handling diverse types of loss functions.
|
Hardware Optimizations of Dense Binary Hyperdimensional Computing:
Rematerialization of Hypervectors, Binarized Bundling, and Combinational
Associative Memory
|
Brain-inspired hyperdimensional (HD) computing models neural activity
patterns of the very size of the brain's circuits with points of a
hyperdimensional space, that is, with hypervectors. Hypervectors are
$D$-dimensional (pseudo)random vectors with independent and identically
distributed (i.i.d.) components constituting ultra-wide holographic words: $D =
10,000$ bits, for instance. At its very core, HD computing manipulates a set of
seed hypervectors to build composite hypervectors representing objects of
interest. It demands memory optimizations with simple operations for an e cient
hardware realization. In this paper, we propose hardware techniques for
optimizations of HD computing, in a synthesizable VHDL library, to enable
co-located implementation of both learning and classification tasks on only a
small portion of Xilinx(R) UltraScale(TM) FPGAs: (1) We propose simple logical
operations to rematerialize the hypervectors on the fly rather than loading
them from memory. These operations massively reduce the memory footprint by
directly computing the composite hypervectors whose individual seed
hypervectors do not need to be stored in memory. (2) Bundling a series of
hypervectors over time requires a multibit counter per every hypervector
component. We instead propose a binarized back-to-back bundling without
requiring any counters. This truly enables on-chip learning with minimal
resources as every hypervector component remains binary over the course of
training to avoid otherwise multibit component. (3) For every classification
event, an associative memory is in charge of finding the closest match between
a set of learned hypervectors and a query hypervector by using a distance
metric. This operator is proportional to [...]
|
Recent Advances in Convolutional Neural Network Acceleration
|
In recent years, convolutional neural networks (CNNs) have shown great
performance in various fields such as image classification, pattern
recognition, and multi-media compression. Two of the feature properties, local
connectivity and weight sharing, can reduce the number of parameters and
increase processing speed during training and inference. However, as the
dimension of data becomes higher and the CNN architecture becomes more
complicated, the end-to-end approach or the combined manner of CNN is
computationally intensive, which becomes limitation to CNN's further
implementation. Therefore, it is necessary and urgent to implement CNN in a
faster way. In this paper, we first summarize the acceleration methods that
contribute to but not limited to CNN by reviewing a broad variety of research
papers. We propose a taxonomy in terms of three levels, i.e.~structure level,
algorithm level, and implementation level, for acceleration methods. We also
analyze the acceleration methods in terms of CNN architecture compression,
algorithm optimization, and hardware-based improvement. At last, we give a
discussion on different perspectives of these acceleration and optimization
methods within each level. The discussion shows that the methods in each level
still have large exploration space. By incorporating such a wide range of
disciplines, we expect to provide a comprehensive reference for researchers who
are interested in CNN acceleration.
|
3D Convolutional Neural Networks for Tumor Segmentation using Long-range
2D Context
|
We present an efficient deep learning approach for the challenging task of
tumor segmentation in multisequence MR images. In recent years, Convolutional
Neural Networks (CNN) have achieved state-of-the-art performances in a large
variety of recognition tasks in medical imaging. Because of the considerable
computational cost of CNNs, large volumes such as MRI are typically processed
by subvolumes, for instance slices (axial, coronal, sagittal) or small 3D
patches. In this paper we introduce a CNN-based model which efficiently
combines the advantages of the short-range 3D context and the long-range 2D
context. To overcome the limitations of specific choices of neural network
architectures, we also propose to merge outputs of several cascaded 2D-3D
models by a voxelwise voting strategy. Furthermore, we propose a network
architecture in which the different MR sequences are processed by separate
subnetworks in order to be more robust to the problem of missing MR sequences.
Finally, a simple and efficient algorithm for training large CNN models is
introduced. We evaluate our method on the public benchmark of the BRATS 2017
challenge on the task of multiclass segmentation of malignant brain tumors. Our
method achieves good performances and produces accurate segmentations with
median Dice scores of 0.918 (whole tumor), 0.883 (tumor core) and 0.854
(enhancing core). Our approach can be naturally applied to various tasks
involving segmentation of lesions or organs.
|
Auto-adaptive Resonance Equalization using Dilated Residual Networks
|
In music and audio production, attenuation of spectral resonances is an
important step towards a technically correct result. In this paper we present a
two-component system to automate the task of resonance equalization. The first
component is a dynamic equalizer that automatically detects resonances and
offers to attenuate them by a user-specified factor. The second component is a
deep neural network that predicts the optimal attenuation factor based on the
windowed audio. The network is trained and validated on empirical data gathered
from an experiment in which sound engineers choose their preferred attenuation
factors for a set of tracks. We test two distinct network architectures for the
predictive model and find that a dilated residual network operating directly on
the audio signal is on a par with a network architecture that requires a prior
audio feature extraction stage. Both architectures predict human-preferred
resonance attenuation factors significantly better than a baseline approach.
|
Understanding the Modeling of Computer Network Delays using Neural
Networks
|
Recent trends in networking are proposing the use of Machine Learning (ML)
techniques for the control and operation of the network. In this context, ML
can be used as a computer network modeling technique to build models that
estimate the network performance. Indeed, network modeling is a central
technique to many networking functions, for instance in the field of
optimization, in which the model is used to search a configuration that
satisfies the target policy. In this paper, we aim to provide an answer to the
following question: Can neural networks accurately model the delay of a
computer network as a function of the input traffic? For this, we assume the
network as a black-box that has as input a traffic matrix and as output delays.
Then we train different neural networks models and evaluate its accuracy under
different fundamental network characteristics: topology, size, traffic
intensity and routing. With this, we aim to have a better understanding of
computer network modeling with neural nets and ultimately provide practical
guidelines on how such models need to be trained.
|
Training Humans and Machines
|
For many years, researchers in psychology, education, statistics, and machine
learning have been developing practical methods to improve learning speed,
retention, and generalizability, and this work has been successful. Many of
these methods are rooted in common underlying principles that seem to drive
learning and overlearning in both humans and machines. I present a review of a
small part of this work to point to potentially novel applications in both
machine and human learning that may be worth exploring.
|
Contrastive Explanations for Reinforcement Learning in terms of Expected
Consequences
|
Machine Learning models become increasingly proficient in complex tasks.
However, even for experts in the field, it can be difficult to understand what
the model learned. This hampers trust and acceptance, and it obstructs the
possibility to correct the model. There is therefore a need for transparency of
machine learning models. The development of transparent classification models
has received much attention, but there are few developments for achieving
transparent Reinforcement Learning (RL) models. In this study we propose a
method that enables a RL agent to explain its behavior in terms of the expected
consequences of state transitions and outcomes. First, we define a translation
of states and actions to a description that is easier to understand for human
users. Second, we developed a procedure that enables the agent to obtain the
consequences of a single action, as well as its entire policy. The method
calculates contrasts between the consequences of a policy derived from a user
query, and of the learned policy of the agent. Third, a format for generating
explanations was constructed. A pilot survey study was conducted to explore
preferences of users for different explanation properties. Results indicate
that human users tend to favor explanations about policy rather than about
single actions.
|
NullaNet: Training Deep Neural Networks for Reduced-Memory-Access
Inference
|
Deep neural networks have been successfully deployed in a wide variety of
applications including computer vision and speech recognition. However,
computational and storage complexity of these models has forced the majority of
computations to be performed on high-end computing platforms or on the cloud.
To cope with computational and storage complexity of these models, this paper
presents a training method that enables a radically different approach for
realization of deep neural networks through Boolean logic minimization. The
aforementioned realization completely removes the energy-hungry step of
accessing memory for obtaining model parameters, consumes about two orders of
magnitude fewer computing resources compared to realizations that use
floatingpoint operations, and has a substantially lower latency.
|
FasTer: Fast Tensor Completion with Nonconvex Regularization
|
Low-rank tensor completion problem aims to recover a tensor from limited
observations, which has many real-world applications. Due to the easy
optimization, the convex overlapping nuclear norm has been popularly used for
tensor completion. However, it over-penalizes top singular values and lead to
biased estimations. In this paper, we propose to use the nonconvex regularizer,
which can less penalize large singular values, instead of the convex one for
tensor completion. However, as the new regularizer is nonconvex and overlapped
with each other, existing algorithms are either too slow or suffer from the
huge memory cost. To address these issues, we develop an efficient and scalable
algorithm, which is based on the proximal average (PA) algorithm, for
real-world problems. Compared with the direct usage of PA algorithm, the
proposed algorithm runs orders faster and needs orders less space. We further
speed up the proposed algorithm with the acceleration technique, and show the
convergence to critical points is still guaranteed. Experimental comparisons of
the proposed approach are made with various other tensor completion approaches.
Empirical results show that the proposed algorithm is very fast and can produce
much better recovery performance.
|
PCNNA: A Photonic Convolutional Neural Network Accelerator
|
Convolutional Neural Networks (CNN) have been the centerpiece of many
applications including but not limited to computer vision, speech processing,
and Natural Language Processing (NLP). However, the computationally expensive
convolution operations impose many challenges to the performance and
scalability of CNNs. In parallel, photonic systems, which are traditionally
employed for data communication, have enjoyed recent popularity for data
processing due to their high bandwidth, low power consumption, and
reconfigurability. Here we propose a Photonic Convolutional Neural Network
Accelerator (PCNNA) as a proof of concept design to speedup the convolution
operation for CNNs. Our design is based on the recently introduced silicon
photonic microring weight banks, which use broadcast-and-weight protocol to
perform Multiply And Accumulate (MAC) operation and move data through layers of
a neural network. Here, we aim to exploit the synergy between the inherent
parallelism of photonics in the form of Wavelength Division Multiplexing (WDM)
and sparsity of connections between input feature maps and kernels in CNNs.
While our full system design offers up to more than 3 orders of magnitude
speedup in execution time, its optical core potentially offers more than 5
order of magnitude speedup compared to state-of-the-art electronic
counterparts.
|
RAIM: Recurrent Attentive and Intensive Model of Multimodal Patient
Monitoring Data
|
With the improvement of medical data capturing, vast amount of continuous
patient monitoring data, e.g., electrocardiogram (ECG), real-time vital signs
and medications, become available for clinical decision support at intensive
care units (ICUs). However, it becomes increasingly challenging to model such
data, due to high density of the monitoring data, heterogeneous data types and
the requirement for interpretable models. Integration of these high-density
monitoring data with the discrete clinical events (including diagnosis,
medications, labs) is challenging but potentially rewarding since richness and
granularity in such multimodal data increase the possibilities for accurate
detection of complex problems and predicting outcomes (e.g., length of stay and
mortality). We propose Recurrent Attentive and Intensive Model (RAIM) for
jointly analyzing continuous monitoring data and discrete clinical events. RAIM
introduces an efficient attention mechanism for continuous monitoring data
(e.g., ECG), which is guided by discrete clinical events (e.g, medication
usage). We apply RAIM in predicting physiological decompensation and length of
stay in those critically ill patients at ICU. With evaluations on MIMIC- III
Waveform Database Matched Subset, we obtain an AUC-ROC score of 90.18% for
predicting decompensation and an accuracy of 86.82% for forecasting length of
stay with our final model, which outperforms our six baseline models.
|
Hierarchical Classification using Binary Data
|
In classification problems, especially those that categorize data into a
large number of classes, the classes often naturally follow a hierarchical
structure. That is, some classes are likely to share similar structures and
features. Those characteristics can be captured by considering a hierarchical
relationship among the class labels. Here, we extend a recent simple
classification approach on binary data in order to efficiently classify
hierarchical data. In certain settings, specifically, when some classes are
significantly easier to identify than others, we showcase computational and
accuracy advantages.
|
Lesion segmentation using U-Net network
|
This paper explains the method used in the segmentation challenge (Task 1) in
the International Skin Imaging Collaboration's (ISIC) Skin Lesion Analysis
Towards Melanoma Detection challenge held in 2018. We have trained a U-Net
network to perform the segmentation. The key elements for the training were
first to adjust the loss function to incorporate unbalanced proportion of
background and second to perform post-processing operation to adjust the
contour of the prediction.
|
Weak in the NEES?: Auto-tuning Kalman Filters with Bayesian Optimization
|
Kalman filters are routinely used for many data fusion applications including
navigation, tracking, and simultaneous localization and mapping problems.
However, significant time and effort is frequently required to tune various
Kalman filter model parameters, e.g. process noise covariance, pre-whitening
filter models for non-white noise, etc. Conventional optimization techniques
for tuning can get stuck in poor local minima and can be expensive to implement
with real sensor data. To address these issues, a new "black box" Bayesian
optimization strategy is developed for automatically tuning Kalman filters. In
this approach, performance is characterized by one of two stochastic objective
functions: normalized estimation error squared (NEES) when ground truth state
models are available, or the normalized innovation error squared (NIS) when
only sensor data is available. By intelligently sampling the parameter space to
both learn and exploit a nonparametric Gaussian process surrogate function for
the NEES/NIS costs, Bayesian optimization can efficiently identify multiple
local minima and provide uncertainty quantification on its results.
|
Supporting Very Large Models using Automatic Dataflow Graph Partitioning
|
This paper presents Tofu, a system that partitions very large DNN models
across multiple GPU devices to reduce per-GPU memory footprint. Tofu is
designed to partition a dataflow graph of fine-grained tensor operators in
order to work transparently with a general-purpose deep learning platform like
MXNet. In order to automatically partition each operator, we propose to
describe the semantics of an operator in a simple language which represents
tensors as lambda functions mapping from tensor coordinates to values. To
optimally partition different operators in a dataflow graph, Tofu uses a
recursive search algorithm that minimizes the total communication cost. Our
experiments on an 8-GPU machine show that Tofu enables the training of very
large CNN and RNN models. It also achieves 25% - 400% speedup over alternative
approaches to train very large models.
|
ClusterNet: 3D Instance Segmentation in RGB-D Images
|
We propose a method for instance-level segmentation that uses RGB-D data as
input and provides detailed information about the location, geometry and number
of individual objects in the scene. This level of understanding is fundamental
for autonomous robots. It enables safe and robust decision-making under the
large uncertainty of the real-world. In our model, we propose to use the first
and second order moments of the object occupancy function to represent an
object instance. We train an hourglass Deep Neural Network (DNN) where each
pixel in the output votes for the 3D position of the corresponding object
center and for the object's size and pose. The final instance segmentation is
achieved through clustering in the space of moments. The object-centric
training loss is defined on the output of the clustering. Our method
outperforms the state-of-the-art instance segmentation method on our
synthesized dataset. We show that our method generalizes well on real-world
data achieving visually better segmentation results.
|
A Structured Perspective of Volumes on Active Learning
|
Active Learning (AL) is a learning task that requires learners interactively
query the labels of the sampled unlabeled instances to minimize the training
outputs with human supervisions. In theoretical study, learners approximate the
version space which covers all possible classification hypothesis into a
bounded convex body and try to shrink the volume of it into a half-space by a
given cut size. However, only the hypersphere with finite VC dimensions has
obtained formal approximation guarantees that hold when the classes of
Euclidean space are separable with a margin. In this paper, we approximate the
version space to a structured {hypersphere} that covers most of the hypotheses,
and then divide the available AL sampling approaches into two kinds of
strategies: Outer Volume Sampling and Inner Volume Sampling. After providing
provable guarantees for the performance of AL in version space, we aggregate
the two kinds of volumes to eliminate their sampling biases via finding the
optimal inscribed hyperspheres in the enclosing space of outer volume. To touch
the version space from Euclidean space, we propose a theoretical bridge called
Volume-based Model that increases the `sampling target-independent'. In
non-linear feature space, spanned by kernel, we use sequential optimization to
globally optimize the original space to a sparse space by halving the size of
the kernel space. Then, the EM (Expectation Maximization) model which returns
the local center helps us to find a local representation. To describe this
process, we propose an easy-to-implement algorithm called Volume-based AL
(VAL).
|
Meta-Learning Priors for Efficient Online Bayesian Regression
|
Gaussian Process (GP) regression has seen widespread use in robotics due to
its generality, simplicity of use, and the utility of Bayesian predictions. The
predominant implementation of GP regression is a nonparameteric kernel-based
approach, as it enables fitting of arbitrary nonlinear functions. However, this
approach suffers from two main drawbacks: (1) it is computationally
inefficient, as computation scales poorly with the number of samples; and (2)
it can be data inefficient, as encoding prior knowledge that can aid the model
through the choice of kernel and associated hyperparameters is often
challenging and unintuitive. In this work, we propose ALPaCA, an algorithm for
efficient Bayesian regression which addresses these issues. ALPaCA uses a
dataset of sample functions to learn a domain-specific, finite-dimensional
feature encoding, as well as a prior over the associated weights, such that
Bayesian linear regression in this feature space yields accurate online
predictions of the posterior predictive density. These features are neural
networks, which are trained via a meta-learning (or "learning-to-learn")
approach. ALPaCA extracts all prior information directly from the dataset,
rather than restricting prior information to the choice of kernel
hyperparameters. Furthermore, by operating in the weight space, it
substantially reduces sample complexity. We investigate the performance of
ALPaCA on two simple regression problems, two simulated robotic systems, and on
a lane-change driving task performed by humans. We find our approach
outperforms kernel-based GP regression, as well as state of the art
meta-learning approaches, thereby providing a promising plug-in tool for many
regression tasks in robotics where scalability and data-efficiency are
important.
|
The Variational Homoencoder: Learning to learn high capacity generative
models from few examples
|
Hierarchical Bayesian methods can unify many related tasks (e.g. k-shot
classification, conditional and unconditional generation) as inference within a
single generative model. However, when this generative model is expressed as a
powerful neural network such as a PixelCNN, we show that existing learning
techniques typically fail to effectively use latent variables. To address this,
we develop a modification of the Variational Autoencoder in which encoded
observations are decoded to new elements from the same class. This technique,
which we call a Variational Homoencoder (VHE), produces a hierarchical latent
variable model which better utilises latent variables. We use the VHE framework
to learn a hierarchical PixelCNN on the Omniglot dataset, which outperforms all
existing models on test set likelihood and achieves strong performance on
one-shot generation and classification tasks. We additionally validate the VHE
on natural images from the YouTube Faces database. Finally, we develop
extensions of the model that apply to richer dataset structures such as
factorial and hierarchical categories.
|
Competitive Inner-Imaging Squeeze and Excitation for Residual Network
|
Residual networks, which use a residual unit to supplement the identity
mappings, enable very deep convolutional architecture to operate well, however,
the residual architecture has been proved to be diverse and redundant, which
may leads to low-efficient modeling. In this work, we propose a competitive
squeeze-excitation (SE) mechanism for the residual network. Re-scaling the
value for each channel in this structure will be determined by the residual and
identity mappings jointly, and this design enables us to expand the meaning of
channel relationship modeling in residual blocks. Modeling of the competition
between residual and identity mappings cause the identity flow to control the
complement of the residual feature maps for itself. Furthermore, we design a
novel inner-imaging competitive SE block to shrink the consumption and re-image
the global features of intermediate network structure, by using the
inner-imaging mechanism, we can model the channel-wise relations with
convolution in spatial. We carry out experiments on the CIFAR, SVHN, and
ImageNet datasets, and the proposed method can challenge state-of-the-art
results.
|
SAAGs: Biased Stochastic Variance Reduction Methods for Large-scale
Learning
|
Stochastic approximation is one of the effective approach to deal with the
large-scale machine learning problems and the recent research has focused on
reduction of variance, caused by the noisy approximations of the gradients. In
this paper, we have proposed novel variants of SAAG-I and II (Stochastic
Average Adjusted Gradient) (Chauhan et al. 2017), called SAAG-III and IV,
respectively. Unlike SAAG-I, starting point is set to average of previous epoch
in SAAG-III, and unlike SAAG-II, the snap point and starting point are set to
average and last iterate of previous epoch in SAAG-IV, respectively. To
determine the step size, we have used Stochastic Backtracking-Armijo line
Search (SBAS) which performs line search only on selected mini-batch of data
points. Since backtracking line search is not suitable for large-scale problems
and the constants used to find the step size, like Lipschitz constant, are not
always available so SBAS could be very effective in such cases. We have
extended SAAGs (I, II, III and IV) to solve non-smooth problems and designed
two update rules for smooth and non-smooth problems. Moreover, our theoretical
results have proved linear convergence of SAAG-IV for all the four combinations
of smoothness and strong-convexity, in expectation. Finally, our experimental
studies have proved the efficacy of proposed methods against the state-of-art
techniques.
|
A Temporal Difference Reinforcement Learning Theory of Emotion: unifying
emotion, cognition and adaptive behavior
|
Emotions are intimately tied to motivation and the adaptation of behavior,
and many animal species show evidence of emotions in their behavior. Therefore,
emotions must be related to powerful mechanisms that aid survival, and,
emotions must be evolutionary continuous phenomena. How and why did emotions
evolve in nature, how do events get emotionally appraised, how do emotions
relate to cognitive complexity, and, how do they impact behavior and learning?
In this article I propose that all emotions are manifestations of reward
processing, in particular Temporal Difference (TD) error assessment.
Reinforcement Learning (RL) is a powerful computational model for the learning
of goal oriented tasks by exploration and feedback. Evidence indicates that
RL-like processes exist in many animal species. Key in the processing of
feedback in RL is the notion of TD error, the assessment of how much better or
worse a situation just became, compared to what was previously expected (or,
the estimated gain or loss of utility - or well-being - resulting from new
evidence). I propose a TDRL Theory of Emotion and discuss its ramifications for
our understanding of emotions in humans, animals and machines, and present
psychological, neurobiological and computational evidence in its support.
|
Space-Time Extension of the MEM Approach for Electromagnetic
Neuroimaging
|
The wavelet Maximum Entropy on the Mean (wMEM) approach to the MEG inverse
problem is revisited and extended to infer brain activity from full space-time
data. The resulting dimensionality increase is tackled using a collection of
techniques , that includes time and space dimension reduction (using
respectively wavelet and spatial filter based reductions), Kronecker product
modeling for covariance matrices, and numerical manipulation of the free energy
directly in matrix form. This leads to a smooth numerical optimization problem
of reasonable dimension, solved using standard approaches. The method is
applied to the MEG inverse problem. Results of a simulation study in the
context of slow wave localization from sleep MEG data are presented and
discussed.
Index Terms: MEG inverse problem, maximum entropy on the mean, wavelet
decomposition, spatial filters, Kronecker covariance factorization, sleep slow
waves.
|
Deep-CLASS at ISIC Machine Learning Challenge 2018
|
This paper reports the method and evaluation results of MedAusbild team for
ISIC challenge task. Since early 2017, our team has worked on melanoma
classification [1][6], and has employed deep learning since beginning of 2018
[7]. Deep learning helps researchers absolutely to treat and detect diseases by
analyzing medical data (e.g., medical images). One of the representative models
among the various deep-learning models is a convolutional neural network (CNN).
Although our team has an experience with segmentation and classification of
benign and malignant skin-lesions, we have participated in the task 3 of ISIC
Challenge 2018 for classification of seven skin diseases, explained in this
paper.
|
Collective Matrix Completion
|
Matrix completion aims to reconstruct a data matrix based on observations of
a small number of its entries. Usually in matrix completion a single matrix is
considered, which can be, for example, a rating matrix in recommendation
system. However, in practical situations, data is often obtained from multiple
sources which results in a collection of matrices rather than a single one. In
this work, we consider the problem of collective matrix completion with
multiple and heterogeneous matrices, which can be count, binary, continuous,
etc. We first investigate the setting where, for each source, the matrix
entries are sampled from an exponential family distribution. Then, we relax the
assumption of exponential family distribution for the noise and we investigate
the distribution-free case. In this setting, we do not assume any specific
model for the observations. The estimation procedures are based on minimizing
the sum of a goodness-of-fit term and the nuclear norm penalization of the
whole collective matrix. We prove that the proposed estimators achieve fast
rates of convergence under the two considered settings and we corroborate our
results with numerical experiments.
|
Uncertainty Modelling in Deep Networks: Forecasting Short and Noisy
Series
|
Deep Learning is a consolidated, state-of-the-art Machine Learning tool to
fit a function when provided with large data sets of examples. However, in
regression tasks, the straightforward application of Deep Learning models
provides a point estimate of the target. In addition, the model does not take
into account the uncertainty of a prediction. This represents a great
limitation for tasks where communicating an erroneous prediction carries a
risk. In this paper we tackle a real-world problem of forecasting impending
financial expenses and incomings of customers, while displaying predictable
monetary amounts on a mobile app. In this context, we investigate if we would
obtain an advantage by applying Deep Learning models with a Heteroscedastic
model of the variance of a network's output. Experimentally, we achieve a
higher accuracy than non-trivial baselines. More importantly, we introduce a
mechanism to discard low-confidence predictions, which means that they will not
be visible to users. This should help enhance the user experience of our
product.
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.