title
stringlengths 7
246
| abstract
stringlengths 6
3.31k
|
|---|---|
Barrier Frank-Wolfe for Marginal Inference | We introduce a globally-convergent algorithm for optimizing the
tree-reweighted (TRW) variational objective over the marginal polytope. The
algorithm is based on the conditional gradient method (Frank-Wolfe) and moves
pseudomarginals within the marginal polytope through repeated maximum a
posteriori (MAP) calls. This modular structure enables us to leverage black-box
MAP solvers (both exact and approximate) for variational inference, and obtains
more accurate results than tree-reweighted algorithms that optimize over the
local consistency relaxation. Theoretically, we bound the sub-optimality for
the proposed algorithm despite the TRW objective having unbounded gradients at
the boundary of the marginal polytope. Empirically, we demonstrate the
increased quality of results found by tightening the relaxation over the
marginal polytope as well as the spanning tree polytope on synthetic and
real-world instances.
|
Diffusion-Convolutional Neural Networks | We present diffusion-convolutional neural networks (DCNNs), a new model for
graph-structured data. Through the introduction of a diffusion-convolution
operation, we show how diffusion-based representations can be learned from
graph-structured data and used as an effective basis for node classification.
DCNNs have several attractive qualities, including a latent representation for
graphical data that is invariant under isomorphism, as well as polynomial-time
prediction and learning that can be represented as tensor operations and
efficiently implemented on the GPU. Through several experiments with real
structured datasets, we demonstrate that DCNNs are able to outperform
probabilistic relational models and kernel-on-graph methods at relational node
classification tasks.
|
Optimal Non-Asymptotic Lower Bound on the Minimax Regret of Learning
with Expert Advice | We prove non-asymptotic lower bounds on the expectation of the maximum of $d$
independent Gaussian variables and the expectation of the maximum of $d$
independent symmetric random walks. Both lower bounds recover the optimal
leading constant in the limit. A simple application of the lower bound for
random walks is an (asymptotically optimal) non-asymptotic lower bound on the
minimax regret of online learning with expert advice.
|
Evaluating Protein-protein Interaction Predictors with a Novel
3-Dimensional Metric | In order for the predicted interactions to be directly adopted by biologists,
the ma- chine learning predictions have to be of high precision, regardless of
recall. This aspect cannot be evaluated or numerically represented well by
traditional metrics like accuracy, ROC, or precision-recall curve. In this
work, we start from the alignment in sensitivity of ROC and recall of
precision-recall curve, and propose an evaluation metric focusing on the
ability of a model to be adopted by biologists. This metric evaluates the
ability of a machine learning algorithm to predict only new interactions,
meanwhile, it eliminates the influence of test dataset. In the experiment of
evaluating different classifiers with a same data set and evaluating the same
predictor with different datasets, our new metric fulfills the evaluation task
of our interest while two widely recognized metrics, ROC and precision-recall
curve fail the tasks for different reasons.
|
Deep Kernel Learning | We introduce scalable deep kernels, which combine the structural properties
of deep learning architectures with the non-parametric flexibility of kernel
methods. Specifically, we transform the inputs of a spectral mixture base
kernel with a deep architecture, using local kernel interpolation, inducing
points, and structure exploiting (Kronecker and Toeplitz) algebra for a
scalable kernel representation. These closed-form kernels can be used as
drop-in replacements for standard kernels, with benefits in expressive power
and scalability. We jointly learn the properties of these kernels through the
marginal likelihood of a Gaussian process. Inference and learning cost $O(n)$
for $n$ training points, and predictions cost $O(1)$ per test point. On a large
and diverse collection of applications, including a dataset with 2 million
examples, we show improved performance over scalable Gaussian processes with
flexible kernel learning models, and stand-alone deep architectures.
|
Active Perceptual Similarity Modeling with Auxiliary Information | Learning a model of perceptual similarity from a collection of objects is a
fundamental task in machine learning underlying numerous applications. A common
way to learn such a model is from relative comparisons in the form of triplets:
responses to queries of the form "Is object a more similar to b than it is to
c?". If no consideration is made in the determination of which queries to ask,
existing similarity learning methods can require a prohibitively large number
of responses. In this work, we consider the problem of actively learning from
triplets -finding which queries are most useful for learning. Different from
previous active triplet learning approaches, we incorporate auxiliary
information into our similarity model and introduce an active learning scheme
to find queries that are informative for quickly learning both the relevant
aspects of auxiliary data and the directly-learned similarity components.
Compared to prior approaches, we show that we can learn just as effectively
with much fewer queries. For evaluation, we introduce a new dataset of
exhaustive triplet comparisons obtained from humans and demonstrate improved
performance for different types of auxiliary information.
|
Efficient Multiscale Gaussian Process Regression using Hierarchical
Clustering | Standard Gaussian Process (GP) regression, a powerful machine learning tool,
is computationally expensive when it is applied to large datasets, and
potentially inaccurate when data points are sparsely distributed in a
high-dimensional feature space. To address these challenges, a new multiscale,
sparsified GP algorithm is formulated, with the goal of application to large
scientific computing datasets. In this approach, the data is partitioned into
clusters and the cluster centers are used to define a reduced training set,
resulting in an improvement over standard GPs in terms of training and
evaluation costs. Further, a hierarchical technique is used to adaptively map
the local covariance representation to the underlying sparsity of the feature
space, leading to improved prediction accuracy when the data distribution is
highly non-uniform. A theoretical investigation of the computational complexity
of the algorithm is presented. The efficacy of this method is then demonstrated
on smooth and discontinuous analytical functions and on data from a direct
numerical simulation of turbulent combustion.
|
Stacked Attention Networks for Image Question Answering | This paper presents stacked attention networks (SANs) that learn to answer
natural language questions from images. SANs use semantic representation of a
question as query to search for the regions in an image that are related to the
answer. We argue that image question answering (QA) often requires multiple
steps of reasoning. Thus, we develop a multiple-layer SAN in which we query an
image multiple times to infer the answer progressively. Experiments conducted
on four image QA data sets demonstrate that the proposed SANs significantly
outperform previous state-of-the-art approaches. The visualization of the
attention layers illustrates the progress that the SAN locates the relevant
visual clues that lead to the answer of the question layer-by-layer.
|
Generation and Comprehension of Unambiguous Object Descriptions | We propose a method that can generate an unambiguous description (known as a
referring expression) of a specific object or region in an image, and which can
also comprehend or interpret such an expression to infer which object is being
described. We show that our method outperforms previous methods that generate
descriptions of objects without taking into account other potentially ambiguous
objects in the scene. Our model is inspired by recent successes of deep
learning methods for image captioning, but while image captioning is difficult
to evaluate, our task allows for easy objective evaluation. We also present a
new large-scale dataset for referring expressions, based on MS-COCO. We have
released the dataset and a toolbox for visualization and evaluation, see
https://github.com/mjhucla/Google_Refexp_toolbox
|
Performance Analysis of Multiclass Support Vector Machine Classification
for Diagnosis of Coronary Heart Diseases | Automatic diagnosis of coronary heart disease helps the doctor to support in
decision making a diagnosis. Coronary heart disease have some types or levels.
Referring to the UCI Repository dataset, it divided into 4 types or levels that
are labeled numbers 1-4 (low, medium, high and serious). The diagnosis models
can be analyzed with multiclass classification approach. One of multiclass
classification approach used, one of which is a support vector machine (SVM).
The SVM use due to strong performance of SVM in binary classification. This
research study multiclass performance classification support vector machine to
diagnose the type or level of coronary heart disease. Coronary heart disease
patient data taken from the UCI Repository. Stages in this study is
preprocessing, which consist of, to normalizing the data, divide the data into
data training and testing. The next stage of multiclass classification and
performance analysis. This study uses multiclass SVM algorithm, namely: Binary
Tree Support Vector Machine (BTSVM), One-Against-One (OAO), One-Against-All
(OAA), Decision Direct Acyclic Graph (DDAG) and Exhaustive Output Error
Correction Code (ECOC). Performance parameter used is recall, precision,
F-measure and Overall accuracy.
|
Review-Level Sentiment Classification with Sentence-Level Polarity
Correction | We propose an effective technique to solving review-level sentiment
classification problem by using sentence-level polarity correction. Our
polarity correction technique takes into account the consistency of the
polarities (positive and negative) of sentences within each product review
before performing the actual machine learning task. While sentences with
inconsistent polarities are removed, sentences with consistent polarities are
used to learn state-of-the-art classifiers. The technique achieved better
results on different types of products reviews and outperforms baseline models
without the correction technique. Experimental results show an average of 82%
F-measure on four different product review domains.
|
Hierarchical Variational Models | Black box variational inference allows researchers to easily prototype and
evaluate an array of models. Recent advances allow such algorithms to scale to
high dimensions. However, a central question remains: How to specify an
expressive variational distribution that maintains efficient computation? To
address this, we develop hierarchical variational models (HVMs). HVMs augment a
variational approximation with a prior on its parameters, which allows it to
capture complex structure for both discrete and continuous latent variables.
The algorithm we develop is black box, can be used for any HVM, and has the
same computational efficiency as the original approximation. We study HVMs on a
variety of deep discrete latent variable models. HVMs generalize other
expressive variational distributions and maintains higher fidelity to the
posterior.
|
Max-Sum Diversification, Monotone Submodular Functions and Semi-metric
Spaces | In many applications such as web-based search, document summarization,
facility location and other applications, the results are preferable to be both
representative and diversified subsets of documents. The goal of this study is
to select a good "quality", bounded-size subset of a given set of items, while
maintaining their diversity relative to a semi-metric distance function. This
problem was first studied by Borodin et al\cite{borodin}, but a crucial
property used throughout their proof is the triangle inequality. In this
modified proof, we want to relax the triangle inequality and relate the
approximation ratio of max-sum diversification problem to the parameter of the
relaxed triangle inequality in the normal form of the problem (i.e., a uniform
matroid) and also in an arbitrary matroid.
|
Towards Structured Deep Neural Network for Automatic Speech Recognition | In this paper we propose the Structured Deep Neural Network (structured DNN)
as a structured and deep learning framework. This approach can learn to find
the best structured object (such as a label sequence) given a structured input
(such as a vector sequence) by globally considering the mapping relationships
between the structures rather than item by item.
When automatic speech recognition is viewed as a special case of such a
structured learning problem, where we have the acoustic vector sequence as the
input and the phoneme label sequence as the output, it becomes possible to
comprehensively learn utterance by utterance as a whole, rather than frame by
frame.
Structured Support Vector Machine (structured SVM) was proposed to perform
ASR with structured learning previously, but limited by the linear nature of
SVM. Here we propose structured DNN to use nonlinear transformations in
multi-layers as a structured and deep learning approach. This approach was
shown to beat structured SVM in preliminary experiments on TIMIT.
|
Algorithmic Stability for Adaptive Data Analysis | Adaptivity is an important feature of data analysis---the choice of questions
to ask about a dataset often depends on previous interactions with the same
dataset. However, statistical validity is typically studied in a nonadaptive
model, where all questions are specified before the dataset is drawn. Recent
work by Dwork et al. (STOC, 2015) and Hardt and Ullman (FOCS, 2014) initiated
the formal study of this problem, and gave the first upper and lower bounds on
the achievable generalization error for adaptive data analysis.
Specifically, suppose there is an unknown distribution $\mathbf{P}$ and a set
of $n$ independent samples $\mathbf{x}$ is drawn from $\mathbf{P}$. We seek an
algorithm that, given $\mathbf{x}$ as input, accurately answers a sequence of
adaptively chosen queries about the unknown distribution $\mathbf{P}$. How many
samples $n$ must we draw from the distribution, as a function of the type of
queries, the number of queries, and the desired level of accuracy?
In this work we make two new contributions:
(i) We give upper bounds on the number of samples $n$ that are needed to
answer statistical queries. The bounds improve and simplify the work of Dwork
et al. (STOC, 2015), and have been applied in subsequent work by those authors
(Science, 2015, NIPS, 2015).
(ii) We prove the first upper bounds on the number of samples required to
answer more general families of queries. These include arbitrary
low-sensitivity queries and an important class of optimization queries.
As in Dwork et al., our algorithms are based on a connection with algorithmic
stability in the form of differential privacy. We extend their work by giving a
quantitatively optimal, more general, and simpler proof of their main theorem
that stability implies low generalization error. We also study weaker stability
guarantees such as bounded KL divergence and total variation distance.
|
Speed learning on the fly | The practical performance of online stochastic gradient descent algorithms is
highly dependent on the chosen step size, which must be tediously hand-tuned in
many applications. The same is true for more advanced variants of stochastic
gradients, such as SAGA, SVRG, or AdaGrad. Here we propose to adapt the step
size by performing a gradient descent on the step size itself, viewing the
whole performance of the learning trajectory as a function of step size.
Importantly, this adaptation can be computed online at little cost, without
having to iterate backward passes over the full data.
|
Sandwiching the marginal likelihood using bidirectional Monte Carlo | Computing the marginal likelihood (ML) of a model requires marginalizing out
all of the parameters and latent variables, a difficult high-dimensional
summation or integration problem. To make matters worse, it is often hard to
measure the accuracy of one's ML estimates. We present bidirectional Monte
Carlo, a technique for obtaining accurate log-ML estimates on data simulated
from a model. This method obtains stochastic lower bounds on the log-ML using
annealed importance sampling or sequential Monte Carlo, and obtains stochastic
upper bounds by running these same algorithms in reverse starting from an exact
posterior sample. The true value can be sandwiched between these two stochastic
bounds with high probability. Using the ground truth log-ML estimates obtained
from our method, we quantitatively evaluate a wide variety of existing ML
estimators on several latent variable models: clustering, a low rank
approximation, and a binary attributes model. These experiments yield insights
into how to accurately estimate marginal likelihoods.
|
Deep Recurrent Neural Networks for Sequential Phenotype Prediction in
Genomics | In analyzing of modern biological data, we are often dealing with ill-posed
problems and missing data, mostly due to high dimensionality and
multicollinearity of the dataset. In this paper, we have proposed a system
based on matrix factorization (MF) and deep recurrent neural networks (DRNNs)
for genotype imputation and phenotype sequences prediction. In order to model
the long-term dependencies of phenotype data, the new Recurrent Linear Units
(ReLU) learning strategy is utilized for the first time. The proposed model is
implemented for parallel processing on central processing units (CPUs) and
graphic processing units (GPUs). Performance of the proposed model is compared
with other training algorithms for learning long-term dependencies as well as
the sparse partial least square (SPLS) method on a set of genotype and
phenotype data with 604 samples, 1980 single-nucleotide polymorphisms (SNPs),
and two traits. The results demonstrate performance of the ReLU training
algorithm in learning long-term dependencies in RNNs.
|
How far can we go without convolution: Improving fully-connected
networks | We propose ways to improve the performance of fully connected networks. We
found that two approaches in particular have a strong effect on performance:
linear bottleneck layers and unsupervised pre-training using autoencoders
without hidden unit biases. We show how both approaches can be related to
improving gradient flow and reducing sparsity in the network. We show that a
fully connected network can yield approximately 70% classification accuracy on
the permutation-invariant CIFAR-10 task, which is much higher than the current
state-of-the-art. By adding deformations to the training data, the fully
connected network achieves 78% accuracy, which is just 10% short of a decent
convolutional network.
|
Batch-normalized Maxout Network in Network | This paper reports a novel deep architecture referred to as Maxout network In
Network (MIN), which can enhance model discriminability and facilitate the
process of information abstraction within the receptive field. The proposed
network adopts the framework of the recently developed Network In Network
structure, which slides a universal approximator, multilayer perceptron (MLP)
with rectifier units, to exact features. Instead of MLP, we employ maxout MLP
to learn a variety of piecewise linear activation functions and to mediate the
problem of vanishing gradients that can occur when using rectifier units.
Moreover, batch normalization is applied to reduce the saturation of maxout
units by pre-conditioning the model and dropout is applied to prevent
overfitting. Finally, average pooling is used in all pooling layers to
regularize maxout MLP in order to facilitate information abstraction in every
receptive field while tolerating the change of object position. Because average
pooling preserves all features in the local patch, the proposed MIN model can
enforce the suppression of irrelevant information during training. Our
experiments demonstrated the state-of-the-art classification performance when
the MIN model was applied to MNIST, CIFAR-10, and CIFAR-100 datasets and
comparable performance for SVHN dataset.
|
A New Relaxation Approach to Normalized Hypergraph Cut | Normalized graph cut (NGC) has become a popular research topic due to its
wide applications in a large variety of areas like machine learning and very
large scale integration (VLSI) circuit design. Most of traditional NGC methods
are based on pairwise relationships (similarities). However, in real-world
applications relationships among the vertices (objects) may be more complex
than pairwise, which are typically represented as hyperedges in hypergraphs.
Thus, normalized hypergraph cut (NHC) has attracted more and more attention.
Existing NHC methods cannot achieve satisfactory performance in real
applications. In this paper, we propose a novel relaxation approach, which is
called relaxed NHC (RNHC), to solve the NHC problem. Our model is defined as an
optimization problem on the Stiefel manifold. To solve this problem, we resort
to the Cayley transformation to devise a feasible learning algorithm.
Experimental results on a set of large hypergraph benchmarks for clustering and
partitioning in VLSI domain show that RNHC can outperform the state-of-the-art
methods.
|
Decomposition Bounds for Marginal MAP | Marginal MAP inference involves making MAP predictions in systems defined
with latent variables or missing information. It is significantly more
difficult than pure marginalization and MAP tasks, for which a large class of
efficient and convergent variational algorithms, such as dual decomposition,
exist. In this work, we generalize dual decomposition to a generic power sum
inference task, which includes marginal MAP, along with pure marginalization
and MAP, as special cases. Our method is based on a block coordinate descent
algorithm on a new convex decomposition bound, that is guaranteed to converge
monotonically, and can be parallelized efficiently. We demonstrate our approach
on marginal MAP queries defined on real-world problems from the UAI approximate
inference challenge, showing that our framework is faster and more reliable
than previous methods.
|
Generating Images from Captions with Attention | Motivated by the recent progress in generative models, we introduce a model
that generates images from natural language descriptions. The proposed model
iteratively draws patches on a canvas, while attending to the relevant words in
the description. After training on Microsoft COCO, we compare our model with
several baseline generative models on image generation and retrieval tasks. We
demonstrate that our model produces higher quality samples than other
approaches and generates images with novel scene compositions corresponding to
previously unseen captions in the dataset.
|
Neural Module Networks | Visual question answering is fundamentally compositional in nature---a
question like "where is the dog?" shares substructure with questions like "what
color is the dog?" and "where is the cat?" This paper seeks to simultaneously
exploit the representational capacity of deep networks and the compositional
linguistic structure of questions. We describe a procedure for constructing and
learning *neural module networks*, which compose collections of jointly-trained
neural "modules" into deep networks for question answering. Our approach
decomposes questions into their linguistic substructures, and uses these
structures to dynamically instantiate modular networks (with reusable
components for recognizing dogs, classifying colors, etc.). The resulting
compound networks are jointly trained. We evaluate our approach on two
challenging datasets for visual question answering, achieving state-of-the-art
results on both the VQA natural image dataset and a new dataset of complex
questions about abstract shapes.
|
Multiple Instance Dictionary Learning using Functions of Multiple
Instances | A multiple instance dictionary learning method using functions of multiple
instances (DL-FUMI) is proposed to address target detection and two-class
classification problems with inaccurate training labels. Given inaccurate
training labels, DL-FUMI learns a set of target dictionary atoms that describe
the most distinctive and representative features of the true positive class as
well as a set of nontarget dictionary atoms that account for the shared
information found in both the positive and negative instances. Experimental
results show that the estimated target dictionary atoms found by DL-FUMI are
more representative prototypes and identify better discriminative features of
the true positive class than existing methods in the literature. DL-FUMI is
shown to have significantly better performance on several target detection and
classification problems as compared to other multiple instance learning (MIL)
dictionary learning algorithms on a variety of MIL problems.
|
Symmetries and control in generative neural nets | We study generative nets which can control and modify observations, after
being trained on real-life datasets. In order to zoom-in on an object, some
spatial, color and other attributes are learned by classifiers in specialized
attention nets. In field-theoretical terms, these learned symmetry statistics
form the gauge group of the data set. Plugging them in the generative layers of
auto-classifiers-encoders (ACE) appears to be the most direct way to
simultaneously: i) generate new observations with arbitrary attributes, from a
given class, ii) describe the low-dimensional manifold encoding the "essence"
of the data, after superfluous attributes are factored out, and iii)
organically control, i.e., move or modify objects within given observations. We
demonstrate the sharp improvement of the generative qualities of shallow ACE,
with added spatial and color symmetry statistics, on the distorted MNIST and
CIFAR10 datasets.
|
Visual Language Modeling on CNN Image Representations | Measuring the naturalness of images is important to generate realistic images
or to detect unnatural regions in images. Additionally, a method to measure
naturalness can be complementary to Convolutional Neural Network (CNN) based
features, which are known to be insensitive to the naturalness of images.
However, most probabilistic image models have insufficient capability of
modeling the complex and abstract naturalness that we feel because they are
built directly on raw image pixels. In this work, we assume that naturalness
can be measured by the predictability on high-level features during eye
movement. Based on this assumption, we propose a novel method to evaluate the
naturalness by building a variant of Recurrent Neural Network Language Models
on pre-trained CNN representations. Our method is applied to two tasks,
demonstrating that 1) using our method as a regularizer enables us to generate
more understandable images from image features than existing approaches, and 2)
unnaturalness maps produced by our method achieve state-of-the-art eye fixation
prediction performance on two well-studied datasets.
|
Neighbourhood NILM: A Big-data Approach to Household Energy
Disaggregation | In this paper, we investigate whether "big-data" is more valuable than
"precise" data for the problem of energy disaggregation: the process of
breaking down aggregate energy usage on a per-appliance basis. Existing
techniques for disaggregation rely on energy metering at a resolution of 1
minute or higher, but most power meters today only provide a reading once per
month, and at most once every 15 minutes. In this paper, we propose a new
technique called Neighbourhood NILM that leverages data from 'neighbouring'
homes to disaggregate energy given only a single energy reading per month. The
key intuition behind our approach is that 'similar' homes have 'similar' energy
consumption on a per-appliance basis. Neighbourhood NILM matches every home
with a set of 'neighbours' that have direct submetering infrastructure, i.e.
power meters on individual circuits or loads. Many such homes already exist.
Then, it estimates the appliance-level energy consumption of the target home to
be the average of its K neighbours. We evaluate this approach using 25 homes
and results show that our approach gives comparable or better disaggregation in
comparison to state-of-the-art accuracy reported in the literature that depend
on manual model training, high frequency power metering, or both. Results show
that Neighbourhood NILM can achieve 83% and 79% accuracy disaggregating fridge
and heating/cooling loads, compared to 74% and 73% for a technique called FHMM.
Furthermore, it achieves up to 64% accuracy on washing machine, dryer,
dishwasher, and lighting loads, which is higher than previously reported
results. Many existing techniques are not able to disaggregate these loads at
all. These results indicate a potentially substantial advantage to installing
submetering infrastructure in a select few homes rather than installing new
high-frequency smart metering infrastructure in all homes.
|
Efficient Construction of Local Parametric Reduced Order Models Using
Machine Learning Techniques | Reduced order models are computationally inexpensive approximations that
capture the important dynamical characteristics of large, high-fidelity
computer models of physical systems. This paper applies machine learning
techniques to improve the design of parametric reduced order models.
Specifically, machine learning is used to develop feasible regions in the
parameter space where the admissible target accuracy is achieved with a
predefined reduced order basis, to construct parametric maps, to chose the best
two already existing bases for a new parameter configuration from accuracy
point of view and to pre-select the optimal dimension of the reduced basis such
as to meet the desired accuracy. By combining available information using bases
concatenation and interpolation as well as high-fidelity solutions
interpolation we are able to build accurate reduced order models associated
with new parameter settings. Promising numerical results with a viscous Burgers
model illustrate the potential of machine learning approaches to help design
better reduced order models.
|
Spectral-Spatial Classification of Hyperspectral Image Using
Autoencoders | Hyperspectral image (HSI) classification is a hot topic in the remote sensing
community. This paper proposes a new framework of spectral-spatial feature
extraction for HSI classification, in which for the first time the concept of
deep learning is introduced. Specifically, the model of autoencoder is
exploited in our framework to extract various kinds of features. First we
verify the eligibility of autoencoder by following classical spectral
information based classification and use autoencoders with different depth to
classify hyperspectral image. Further in the proposed framework, we combine PCA
on spectral dimension and autoencoder on the other two spatial dimensions to
extract spectral-spatial information for classification. The experimental
results show that this framework achieves the highest classification accuracy
among all methods, and outperforms classical classifiers such as SVM and
PCA-based SVM.
|
Reducing the Training Time of Neural Networks by Partitioning | This paper presents a new method for pre-training neural networks that can
decrease the total training time for a neural network while maintaining the
final performance, which motivates its use on deep neural networks. By
partitioning the training task in multiple training subtasks with sub-models,
which can be performed independently and in parallel, it is shown that the size
of the sub-models reduces almost quadratically with the number of subtasks
created, quickly scaling down the sub-models used for the pre-training. The
sub-models are then merged to provide a pre-trained initial set of weights for
the original model. The proposed method is independent of the other aspects of
the training, such as architecture of the neural network, training method, and
objective, making it compatible with a wide range of existing approaches. The
speedup without loss of performance is validated experimentally on MNIST and on
CIFAR10 data sets, also showing that even performing the subtasks sequentially
can decrease the training time. Moreover, we show that larger models may
present higher speedups and conjecture about the benefits of the method in
distributed learning systems.
|
Learning with a Strong Adversary | The robustness of neural networks to intended perturbations has recently
attracted significant attention. In this paper, we propose a new method,
\emph{learning with a strong adversary}, that learns robust classifiers from
supervised data. The proposed method takes finding adversarial examples as an
intermediate step. A new and simple way of finding adversarial examples is
presented and experimentally shown to be efficient. Experimental results
demonstrate that resulting learning method greatly improves the robustness of
the classification models produced.
|
Tiny Descriptors for Image Retrieval with Unsupervised Triplet Hashing | A typical image retrieval pipeline starts with the comparison of global
descriptors from a large database to find a short list of candidate matches. A
good image descriptor is key to the retrieval pipeline and should reconcile two
contradictory requirements: providing recall rates as high as possible and
being as compact as possible for fast matching. Following the recent successes
of Deep Convolutional Neural Networks (DCNN) for large scale image
classification, descriptors extracted from DCNNs are increasingly used in place
of the traditional hand crafted descriptors such as Fisher Vectors (FV) with
better retrieval performances. Nevertheless, the dimensionality of a typical
DCNN descriptor --extracted either from the visual feature pyramid or the
fully-connected layers-- remains quite high at several thousands of scalar
values. In this paper, we propose Unsupervised Triplet Hashing (UTH), a fully
unsupervised method to compute extremely compact binary hashes --in the 32-256
bits range-- from high-dimensional global descriptors. UTH consists of two
successive deep learning steps. First, Stacked Restricted Boltzmann Machines
(SRBM), a type of unsupervised deep neural nets, are used to learn binary
embedding functions able to bring the descriptor size down to the desired
bitrate. SRBMs are typically able to ensure a very high compression rate at the
expense of loosing some desirable metric properties of the original DCNN
descriptor space. Then, triplet networks, a rank learning scheme based on
weight sharing nets is used to fine-tune the binary embedding functions to
retain as much as possible of the useful metric properties of the original
space. A thorough empirical evaluation conducted on multiple publicly available
dataset using DCNN descriptors shows that our method is able to significantly
outperform state-of-the-art unsupervised schemes in the target bit range.
|
The CTU Prague Relational Learning Repository | The aim of the CTU Prague Relational Learning Repository is to support
machine learning research with multi-relational data. The repository currently
contains 50 SQL databases hosted on a public MySQL server located at
relational.fit.cvut.cz. A searchable meta-database provides metadata (e.g., the
number of tables in the database, the number of rows and columns in the tables,
the number of foreign key constraints between tables).
|
Semi-supervised Tuning from Temporal Coherence | Recent works demonstrated the usefulness of temporal coherence to regularize
supervised training or to learn invariant features with deep architectures. In
particular, enforcing smooth output changes while presenting temporally-closed
frames from video sequences, proved to be an effective strategy. In this paper
we prove the efficacy of temporal coherence for semi-supervised incremental
tuning. We show that a deep architecture, just mildly trained in a supervised
manner, can progressively improve its classification accuracy, if exposed to
video sequences of unlabeled data. The extent to which, in some cases, a
semi-supervised tuning allows to improve classification accuracy (approaching
the supervised one) is somewhat surprising. A number of control experiments
pointed out the fundamental role of temporal coherence.
|
Sliced Wasserstein Kernels for Probability Distributions | Optimal transport distances, otherwise known as Wasserstein distances, have
recently drawn ample attention in computer vision and machine learning as a
powerful discrepancy measure for probability distributions. The recent
developments on alternative formulations of the optimal transport have allowed
for faster solutions to the problem and has revamped its practical applications
in machine learning. In this paper, we exploit the widely used kernel methods
and provide a family of provably positive definite kernels based on the Sliced
Wasserstein distance and demonstrate the benefits of these kernels in a variety
of learning tasks. Our work provides a new perspective on the application of
optimal transport flavored distances through kernel methods in machine learning
tasks.
|
Label Efficient Learning by Exploiting Multi-class Output Codes | We present a new perspective on the popular multi-class algorithmic
techniques of one-vs-all and error correcting output codes. Rather than
studying the behavior of these techniques for supervised learning, we establish
a connection between the success of these methods and the existence of
label-efficient learning procedures. We show that in both the realizable and
agnostic cases, if output codes are successful at learning from labeled data,
they implicitly assume structure on how the classes are related. By making that
structure explicit, we design learning algorithms to recover the classes with
low label complexity. We provide results for the commonly studied cases of
one-vs-all learning and when the codewords of the classes are well separated.
We additionally consider the more challenging case where the codewords are not
well separated, but satisfy a boundary features condition that captures the
natural intuition that every bit of the codewords should be significant.
|
Learning Communities in the Presence of Errors | We study the problem of learning communities in the presence of modeling
errors and give robust recovery algorithms for the Stochastic Block Model
(SBM). This model, which is also known as the Planted Partition Model, is
widely used for community detection and graph partitioning in various fields,
including machine learning, statistics, and social sciences. Many algorithms
exist for learning communities in the Stochastic Block Model, but they do not
work well in the presence of errors.
In this paper, we initiate the study of robust algorithms for partial
recovery in SBM with modeling errors or noise. We consider graphs generated
according to the Stochastic Block Model and then modified by an adversary. We
allow two types of adversarial errors, Feige---Kilian or monotone errors, and
edge outlier errors. Mossel, Neeman and Sly (STOC 2015) posed an open question
about whether an almost exact recovery is possible when the adversary is
allowed to add $o(n)$ edges. Our work answers this question affirmatively even
in the case of $k>2$ communities.
We then show that our algorithms work not only when the instances come from
SBM, but also work when the instances come from any distribution of graphs that
is $\epsilon m$ close to SBM in the Kullback---Leibler divergence. This result
also works in the presence of adversarial errors. Finally, we present almost
tight lower bounds for two communities.
|
A Hierarchical Spectral Method for Extreme Classification | Extreme classification problems are multiclass and multilabel classification
problems where the number of outputs is so large that straightforward
strategies are neither statistically nor computationally viable. One strategy
for dealing with the computational burden is via a tree decomposition of the
output space. While this typically leads to training and inference that scales
sublinearly with the number of outputs, it also results in reduced statistical
performance. In this work, we identify two shortcomings of tree decomposition
methods, and describe two heuristic mitigations. We compose these with an
eigenvalue technique for constructing the tree. The end result is a
computationally efficient algorithm that provides good statistical performance
on several extreme data sets.
|
Anchored Discrete Factor Analysis | We present a semi-supervised learning algorithm for learning discrete factor
analysis models with arbitrary structure on the latent variables. Our algorithm
assumes that every latent variable has an "anchor", an observed variable with
only that latent variable as its parent. Given such anchors, we show that it is
possible to consistently recover moments of the latent variables and use these
moments to learn complete models. We also introduce a new technique for
improving the robustness of method-of-moment algorithms by optimizing over the
marginal polytope or its relaxations. We evaluate our algorithm using two
real-world tasks, tag prediction on questions from the Stack Overflow website
and medical diagnosis in an emergency department.
|
Visual7W: Grounded Question Answering in Images | We have seen great progress in basic perceptual tasks such as object
recognition and detection. However, AI models still fail to match humans in
high-level vision tasks due to the lack of capacities for deeper reasoning.
Recently the new task of visual question answering (QA) has been proposed to
evaluate a model's capacity for deep image understanding. Previous works have
established a loose, global association between QA sentences and images.
However, many questions and answers, in practice, relate to local regions in
the images. We establish a semantic link between textual descriptions and image
regions by object-level grounding. It enables a new type of QA with visual
answers, in addition to textual answers used in previous work. We study the
visual QA tasks in a grounded setting with a large collection of 7W
multiple-choice QA pairs. Furthermore, we evaluate human performance and
several baseline models on the QA tasks. Finally, we propose a novel LSTM model
with spatial attention to tackle the 7W QA tasks.
|
Hierarchical Latent Semantic Mapping for Automated Topic Generation | Much of information sits in an unprecedented amount of text data. Managing
allocation of these large scale text data is an important problem for many
areas. Topic modeling performs well in this problem. The traditional generative
models (PLSA,LDA) are the state-of-the-art approaches in topic modeling and
most recent research on topic generation has been focusing on improving or
extending these models. However, results of traditional generative models are
sensitive to the number of topics K, which must be specified manually. The
problem of generating topics from corpus resembles community detection in
networks. Many effective algorithms can automatically detect communities from
networks without a manually specified number of the communities. Inspired by
these algorithms, in this paper, we propose a novel method named Hierarchical
Latent Semantic Mapping (HLSM), which automatically generates topics from
corpus. HLSM calculates the association between each pair of words in the
latent topic space, then constructs a unipartite network of words with this
association and hierarchically generates topics from this network. We apply
HLSM to several document collections and the experimental comparisons against
several state-of-the-art approaches demonstrate the promising performance.
|
Federated Optimization:Distributed Optimization Beyond the Datacenter | We introduce a new and increasingly relevant setting for distributed
optimization in machine learning, where the data defining the optimization are
distributed (unevenly) over an extremely large number of \nodes, but the goal
remains to train a high-quality centralized model. We refer to this setting as
Federated Optimization. In this setting, communication efficiency is of utmost
importance.
A motivating example for federated optimization arises when we keep the
training data locally on users' mobile devices rather than logging it to a data
center for training. Instead, the mobile devices are used as nodes performing
computation on their local data in order to update a global model. We suppose
that we have an extremely large number of devices in our network, each of which
has only a tiny fraction of data available totally; in particular, we expect
the number of data points available locally to be much smaller than the number
of devices. Additionally, since different users generate data with different
patterns, we assume that no device has a representative sample of the overall
distribution.
We show that existing algorithms are not suitable for this setting, and
propose a new algorithm which shows encouraging experimental results. This work
also sets a path for future research needed in the context of federated
optimization.
|
DataGrinder: Fast, Accurate, Fully non-Parametric Classification
Approach Using 2D Convex Hulls | It has been a long time, since data mining technologies have made their ways
to the field of data management. Classification is one of the most important
data mining tasks for label prediction, categorization of objects into groups,
advertisement and data management. In this paper, we focus on the standard
classification problem which is predicting unknown labels in Euclidean space.
Most efforts in Machine Learning communities are devoted to methods that use
probabilistic algorithms which are heavy on Calculus and Linear Algebra. Most
of these techniques have scalability issues for big data, and are hardly
parallelizable if they are to maintain their high accuracies in their standard
form. Sampling is a new direction for improving scalability, using many small
parallel classifiers. In this paper, rather than conventional sampling methods,
we focus on a discrete classification algorithm with O(n) expected running
time. Our approach performs a similar task as sampling methods. However, we use
column-wise sampling of data, rather than the row-wise sampling used in the
literature. In either case, our algorithm is completely deterministic. Our
algorithm, proposes a way of combining 2D convex hulls in order to achieve high
classification accuracy as well as scalability in the same time. First, we
thoroughly describe and prove our O(n) algorithm for finding the convex hull of
a point set in 2D. Then, we show with experiments our classifier model built
based on this idea is very competitive compared with existing sophisticated
classification algorithms included in commercial statistical applications such
as MATLAB.
|
The Fourier Transform of Poisson Multinomial Distributions and its
Algorithmic Applications | An $(n, k)$-Poisson Multinomial Distribution (PMD) is a random variable of
the form $X = \sum_{i=1}^n X_i$, where the $X_i$'s are independent random
vectors supported on the set of standard basis vectors in $\mathbb{R}^k.$ In
this paper, we obtain a refined structural understanding of PMDs by analyzing
their Fourier transform. As our core structural result, we prove that the
Fourier transform of PMDs is {\em approximately sparse}, i.e., roughly
speaking, its $L_1$-norm is small outside a small set. By building on this
result, we obtain the following applications:
{\bf Learning Theory.} We design the first computationally efficient learning
algorithm for PMDs with respect to the total variation distance. Our algorithm
learns an arbitrary $(n, k)$-PMD within variation distance $\epsilon$ using a
near-optimal sample size of $\widetilde{O}_k(1/\epsilon^2),$ and runs in time
$\widetilde{O}_k(1/\epsilon^2) \cdot \log n.$ Previously, no algorithm with a
$\mathrm{poly}(1/\epsilon)$ runtime was known, even for $k=3.$
{\bf Game Theory.} We give the first efficient polynomial-time approximation
scheme (EPTAS) for computing Nash equilibria in anonymous games. For normalized
anonymous games with $n$ players and $k$ strategies, our algorithm computes a
well-supported $\epsilon$-Nash equilibrium in time $n^{O(k^3)} \cdot
(k/\epsilon)^{O(k^3\log(k/\epsilon)/\log\log(k/\epsilon))^{k-1}}.$ The best
previous algorithm for this problem had running time $n^{(f(k)/\epsilon)^k},$
where $f(k) = \Omega(k^{k^2})$, for any $k>2.$
{\bf Statistics.} We prove a multivariate central limit theorem (CLT) that
relates an arbitrary PMD to a discretized multivariate Gaussian with the same
mean and covariance, in total variation distance. Our new CLT strengthens the
CLT of Valiant and Valiant by completely removing the dependence on $n$ in the
error bound.
|
A Size-Free CLT for Poisson Multinomials and its Applications | An $(n,k)$-Poisson Multinomial Distribution (PMD) is the distribution of the
sum of $n$ independent random vectors supported on the set ${\cal
B}_k=\{e_1,\ldots,e_k\}$ of standard basis vectors in $\mathbb{R}^k$. We show
that any $(n,k)$-PMD is ${\rm poly}\left({k\over \sigma}\right)$-close in total
variation distance to the (appropriately discretized) multi-dimensional
Gaussian with the same first two moments, removing the dependence on $n$ from
the Central Limit Theorem of Valiant and Valiant. Interestingly, our CLT is
obtained by bootstrapping the Valiant-Valiant CLT itself through the structural
characterization of PMDs shown in recent work by Daskalakis, Kamath, and
Tzamos. In turn, our stronger CLT can be leveraged to obtain an efficient PTAS
for approximate Nash equilibria in anonymous games, significantly improving the
state of the art, and matching qualitatively the running time dependence on $n$
and $1/\varepsilon$ of the best known algorithm for two-strategy anonymous
games. Our new CLT also enables the construction of covers for the set of
$(n,k)$-PMDs, which are proper and whose size is shown to be essentially
optimal. Our cover construction combines our CLT with the Shapley-Folkman
theorem and recent sparsification results for Laplacian matrices by Batson,
Spielman, and Srivastava. Our cover size lower bound is based on an algebraic
geometric construction. Finally, leveraging the structural properties of the
Fourier spectrum of PMDs we show that these distributions can be learned from
$O_k(1/\varepsilon^2)$ samples in ${\rm poly}_k(1/\varepsilon)$-time, removing
the quasi-polynomial dependence of the running time on $1/\varepsilon$ from the
algorithm of Daskalakis, Kamath, and Tzamos.
|
Unifying distillation and privileged information | Distillation (Hinton et al., 2015) and privileged information (Vapnik &
Izmailov, 2015) are two techniques that enable machines to learn from other
machines. This paper unifies these two techniques into generalized
distillation, a framework to learn from multiple machines and data
representations. We provide theoretical and causal insight about the inner
workings of generalized distillation, extend it to unsupervised, semisupervised
and multitask learning scenarios, and illustrate its efficacy on a variety of
numerical simulations on both synthetic and real-world data.
|
Learning to Diagnose with LSTM Recurrent Neural Networks | Clinical medical data, especially in the intensive care unit (ICU), consist
of multivariate time series of observations. For each patient visit (or
episode), sensor data and lab test results are recorded in the patient's
Electronic Health Record (EHR). While potentially containing a wealth of
insights, the data is difficult to mine effectively, owing to varying length,
irregular sampling and missing data. Recurrent Neural Networks (RNNs),
particularly those using Long Short-Term Memory (LSTM) hidden units, are
powerful and increasingly popular models for learning from sequence data. They
effectively model varying length sequences and capture long range dependencies.
We present the first study to empirically evaluate the ability of LSTMs to
recognize patterns in multivariate time series of clinical measurements.
Specifically, we consider multilabel classification of diagnoses, training a
model to classify 128 diagnoses given 13 frequently but irregularly sampled
clinical measurements. First, we establish the effectiveness of a simple LSTM
network for modeling clinical data. Then we demonstrate a straightforward and
effective training strategy in which we replicate targets at each sequence
step. Trained only on raw time series, our models outperform several strong
baselines, including a multilayer perceptron trained on hand-engineered
features.
|
Generative Concatenative Nets Jointly Learn to Write and Classify
Reviews | A recommender system's basic task is to estimate how users will respond to
unseen items. This is typically modeled in terms of how a user might rate a
product, but here we aim to extend such approaches to model how a user would
write about the product. To do so, we design a character-level Recurrent Neural
Network (RNN) that generates personalized product reviews. The network
convincingly learns styles and opinions of nearly 1000 distinct authors, using
a large corpus of reviews from BeerAdvocate.com. It also tailors reviews to
describe specific items, categories, and star ratings. Using a simple input
replication strategy, the Generative Concatenative Network (GCN) preserves the
signal of static auxiliary inputs across wide sequence intervals. Without any
additional training, the generative model can classify reviews, identifying the
author of the review, the product category, and the sentiment (rating), with
remarkable accuracy. Our evaluation shows the GCN captures complex dynamics in
text, such as the effect of negation, misspellings, slang, and large
vocabularies gracefully absent any machinery explicitly dedicated to the
purpose.
|
Online Principal Component Analysis in High Dimension: Which Algorithm
to Choose? | In the current context of data explosion, online techniques that do not
require storing all data in memory are indispensable to routinely perform tasks
like principal component analysis (PCA). Recursive algorithms that update the
PCA with each new observation have been studied in various fields of research
and found wide applications in industrial monitoring, computer vision,
astronomy, and latent semantic indexing, among others. This work provides
guidance for selecting an online PCA algorithm in practice. We present the main
approaches to online PCA, namely, perturbation techniques, incremental methods,
and stochastic optimization, and compare their statistical accuracy,
computation time, and memory requirements using artificial and real data.
Extensions to missing data and to functional data are discussed. All studied
algorithms are available in the R package onlinePCA on CRAN.
|
Universum Prescription: Regularization using Unlabeled Data | This paper shows that simply prescribing "none of the above" labels to
unlabeled data has a beneficial regularization effect to supervised learning.
We call it universum prescription by the fact that the prescribed labels cannot
be one of the supervised labels. In spite of its simplicity, universum
prescription obtained competitive results in training deep convolutional
networks for CIFAR-10, CIFAR-100, STL-10 and ImageNet datasets. A qualitative
justification of these approaches using Rademacher complexity is presented. The
effect of a regularization parameter -- probability of sampling from unlabeled
data -- is also studied empirically.
|
Doubly Robust Off-policy Value Evaluation for Reinforcement Learning | We study the problem of off-policy value evaluation in reinforcement learning
(RL), where one aims to estimate the value of a new policy based on data
collected by a different policy. This problem is often a critical step when
applying RL in real-world problems. Despite its importance, existing general
methods either have uncontrolled bias or suffer high variance. In this work, we
extend the doubly robust estimator for bandits to sequential decision-making
problems, which gets the best of both worlds: it is guaranteed to be unbiased
and can have a much lower variance than the popular importance sampling
estimators. We demonstrate the estimator's accuracy in several benchmark
problems, and illustrate its use as a subroutine in safe policy improvement. We
also provide theoretical results on the hardness of the problem, and show that
our estimator can match the lower bound in certain scenarios.
|
Grounding of Textual Phrases in Images by Reconstruction | Grounding (i.e. localizing) arbitrary, free-form textual phrases in visual
content is a challenging problem with many applications for human-computer
interaction and image-text reference resolution. Few datasets provide the
ground truth spatial localization of phrases, thus it is desirable to learn
from data with no or little grounding supervision. We propose a novel approach
which learns grounding by reconstructing a given phrase using an attention
mechanism, which can be either latent or optimized directly. During training
our approach encodes the phrase using a recurrent network language model and
then learns to attend to the relevant image region in order to reconstruct the
input phrase. At test time, the correct attention, i.e., the grounding, is
evaluated. If grounding supervision is available it can be directly applied via
a loss over the attention mechanism. We demonstrate the effectiveness of our
approach on the Flickr 30k Entities and ReferItGame datasets with different
levels of supervision, ranging from no supervision over partial supervision to
full supervision. Our supervised variant improves by a large margin over the
state-of-the-art on both datasets.
|
Random Multi-Constraint Projection: Stochastic Gradient Methods for
Convex Optimization with Many Constraints | Consider convex optimization problems subject to a large number of
constraints. We focus on stochastic problems in which the objective takes the
form of expected values and the feasible set is the intersection of a large
number of convex sets. We propose a class of algorithms that perform both
stochastic gradient descent and random feasibility updates simultaneously. At
every iteration, the algorithms sample a number of projection points onto a
randomly selected small subsets of all constraints. Three feasibility update
schemes are considered: averaging over random projected points, projecting onto
the most distant sample, projecting onto a special polyhedral set constructed
based on sample points. We prove the almost sure convergence of these
algorithms, and analyze the iterates' feasibility error and optimality error,
respectively. We provide new convergence rate benchmarks for stochastic
first-order optimization with many constraints. The rate analysis and numerical
experiments reveal that the algorithm using the polyhedral-set projection
scheme is the most efficient one within known algorithms.
|
Sparse Learning for Large-scale and High-dimensional Data: A Randomized
Convex-concave Optimization Approach | In this paper, we develop a randomized algorithm and theory for learning a
sparse model from large-scale and high-dimensional data, which is usually
formulated as an empirical risk minimization problem with a sparsity-inducing
regularizer. Under the assumption that there exists a (approximately) sparse
solution with high classification accuracy, we argue that the dual solution is
also sparse or approximately sparse. The fact that both primal and dual
solutions are sparse motivates us to develop a randomized approach for a
general convex-concave optimization problem. Specifically, the proposed
approach combines the strength of random projection with that of sparse
learning: it utilizes random projection to reduce the dimensionality, and
introduces $\ell_1$-norm regularization to alleviate the approximation error
caused by random projection. Theoretical analysis shows that under favored
conditions, the randomized algorithm can accurately recover the optimal
solutions to the convex-concave optimization problem (i.e., recover both the
primal and dual solutions).
|
Improving performance of recurrent neural network with relu nonlinearity | In recent years significant progress has been made in successfully training
recurrent neural networks (RNNs) on sequence learning problems involving long
range temporal dependencies. The progress has been made on three fronts: (a)
Algorithmic improvements involving sophisticated optimization techniques, (b)
network design involving complex hidden layer nodes and specialized recurrent
layer connections and (c) weight initialization methods. In this paper, we
focus on recently proposed weight initialization with identity matrix for the
recurrent weights in a RNN. This initialization is specifically proposed for
hidden nodes with Rectified Linear Unit (ReLU) non linearity. We offer a simple
dynamical systems perspective on weight initialization process, which allows us
to propose a modified weight initialization strategy. We show that this
initialization technique leads to successfully training RNNs composed of ReLUs.
We demonstrate that our proposal produces comparable or better solution for
three toy problems involving long range temporal structure: the addition
problem, the multiplication problem and the MNIST classification problem using
sequence of pixels. In addition, we present results for a benchmark action
recognition problem.
|
On the Optimal Sample Complexity for Best Arm Identification | We study the best arm identification (BEST-1-ARM) problem, which is defined
as follows. We are given $n$ stochastic bandit arms. The $i$th arm has a reward
distribution $D_i$ with an unknown mean $\mu_{i}$. Upon each play of the $i$th
arm, we can get a reward, sampled i.i.d. from $D_i$. We would like to identify
the arm with the largest mean with probability at least $1-\delta$, using as
few samples as possible. We provide a nontrivial algorithm for BEST-1-ARM,
which improves upon several prior upper bounds on the same problem. We also
study an important special case where there are only two arms, which we call
the sign problem. We provide a new lower bound of sign, simplifying and
significantly extending a classical result by Farrell in 1964, with a
completely new proof. Using the new lower bound for sign, we obtain the first
lower bound for BEST-1-ARM that goes beyond the classic Mannor-Tsitsiklis lower
bound, by an interesting reduction from Sign to BEST-1-ARM. We propose an
interesting conjecture concerning the optimal sample complexity of BEST-1-ARM
from the perspective of instance-wise optimality.
|
Towards Vision-Based Deep Reinforcement Learning for Robotic Motion
Control | This paper introduces a machine learning based system for controlling a
robotic manipulator with visual perception only. The capability to autonomously
learn robot controllers solely from raw-pixel images and without any prior
knowledge of configuration is shown for the first time. We build upon the
success of recent deep reinforcement learning and develop a system for learning
target reaching with a three-joint robot manipulator using external visual
observation. A Deep Q Network (DQN) was demonstrated to perform target reaching
after training in simulation. Transferring the network to real hardware and
real observation in a naive approach failed, but experiments show that the
network works when replacing camera images with synthetic images.
|
Learning Nonparametric Forest Graphical Models with Prior Information | We present a framework for incorporating prior information into nonparametric
estimation of graphical models. To avoid distributional assumptions, we
restrict the graph to be a forest and build on the work of forest density
estimation (FDE). We reformulate the FDE approach from a Bayesian perspective,
and introduce prior distributions on the graphs. As two concrete examples, we
apply this framework to estimating scale-free graphs and learning multiple
graphs with similar structures. The resulting algorithms are equivalent to
finding a maximum spanning tree of a weighted graph with a penalty term on the
connectivity pattern of the graph. We solve the optimization problem via a
minorize-maximization procedure with Kruskal's algorithm. Simulations show that
the proposed methods outperform competing parametric methods, and are robust to
the true data distribution. They also lead to improvement in predictive power
and interpretability in two real data sets.
|
Characterizing Concept Drift | Most machine learning models are static, but the world is dynamic, and
increasing online deployment of learned models gives increasing urgency to the
development of efficient and effective mechanisms to address learning in the
context of non-stationary distributions, or as it is commonly called concept
drift. However, the key issue of characterizing the different types of drift
that can occur has not previously been subjected to rigorous definition and
analysis. In particular, while some qualitative drift categorizations have been
proposed, few have been formally defined, and the quantitative descriptions
required for precise and objective understanding of learner performance have
not existed. We present the first comprehensive framework for quantitative
analysis of drift. This supports the development of the first comprehensive set
of formal definitions of types of concept drift. The formal definitions clarify
ambiguities and identify gaps in previous definitions, giving rise to a new
comprehensive taxonomy of concept drift types and a solid foundation for
research into mechanisms to detect and address concept drift.
|
Feature Learning based Deep Supervised Hashing with Pairwise Labels | Recent years have witnessed wide application of hashing for large-scale image
retrieval. However, most existing hashing methods are based on hand-crafted
features which might not be optimally compatible with the hashing procedure.
Recently, deep hashing methods have been proposed to perform simultaneous
feature learning and hash-code learning with deep neural networks, which have
shown better performance than traditional hashing methods with hand-crafted
features. Most of these deep hashing methods are supervised whose supervised
information is given with triplet labels. For another common application
scenario with pairwise labels, there have not existed methods for simultaneous
feature learning and hash-code learning. In this paper, we propose a novel deep
hashing method, called deep pairwise-supervised hashing(DPSH), to perform
simultaneous feature learning and hash-code learning for applications with
pairwise labels. Experiments on real datasets show that our DPSH method can
outperform other methods to achieve the state-of-the-art performance in image
retrieval applications.
|
Learning Human Identity from Motion Patterns | We present a large-scale study exploring the capability of temporal deep
neural networks to interpret natural human kinematics and introduce the first
method for active biometric authentication with mobile inertial sensors. At
Google, we have created a first-of-its-kind dataset of human movements,
passively collected by 1500 volunteers using their smartphones daily over
several months. We (1) compare several neural architectures for efficient
learning of temporal multi-modal data representations, (2) propose an optimized
shift-invariant dense convolutional mechanism (DCWRNN), and (3) incorporate the
discriminatively-trained dynamic features in a probabilistic generative
framework taking into account temporal characteristics. Our results demonstrate
that human kinematics convey important information about user identity and can
serve as a valuable component of multi-modal authentication systems.
|
Bayesian Analysis of Dynamic Linear Topic Models | In dynamic topic modeling, the proportional contribution of a topic to a
document depends on the temporal dynamics of that topic's overall prevalence in
the corpus. We extend the Dynamic Topic Model of Blei and Lafferty (2006) by
explicitly modeling document level topic proportions with covariates and
dynamic structure that includes polynomial trends and periodicity. A Markov
Chain Monte Carlo (MCMC) algorithm that utilizes Polya-Gamma data augmentation
is developed for posterior inference. Conditional independencies in the model
and sampling are made explicit, and our MCMC algorithm is parallelized where
possible to allow for inference in large corpora. To address computational
bottlenecks associated with Polya-Gamma sampling, we appeal to the Central
Limit Theorem to develop a Gaussian approximation to the Polya-Gamma random
variable. This approximation is fast and reliable for parameter values relevant
in the text mining domain. Our model and inference algorithm are validated with
multiple simulation examples, and we consider the application of modeling
trends in PubMed abstracts. We demonstrate that sharing information across
documents is critical for accurately estimating document-specific topic
proportions. We also show that explicitly modeling polynomial and periodic
behavior improves our ability to predict topic prevalence at future time
points.
|
Document Context Language Models | Text documents are structured on multiple levels of detail: individual words
are related by syntax, but larger units of text are related by discourse
structure. Existing language models generally fail to account for discourse
structure, but it is crucial if we are to have language models that reward
coherence and generate coherent texts. We present and empirically evaluate a
set of multi-level recurrent neural network language models, called
Document-Context Language Models (DCLM), which incorporate contextual
information both within and beyond the sentence. In comparison with word-level
recurrent neural network language models, the DCLM models obtain slightly
better predictive likelihoods, and considerably better assessments of document
coherence.
|
Representational Distance Learning for Deep Neural Networks | Deep neural networks (DNNs) provide useful models of visual representational
transformations. We present a method that enables a DNN (student) to learn from
the internal representational spaces of a reference model (teacher), which
could be another DNN or, in the future, a biological brain. Representational
spaces of the student and the teacher are characterized by representational
distance matrices (RDMs). We propose representational distance learning (RDL),
a stochastic gradient descent method that drives the RDMs of the student to
approximate the RDMs of the teacher. We demonstrate that RDL is competitive
with other transfer learning techniques for two publicly available benchmark
computer vision datasets (MNIST and CIFAR-100), while allowing for
architectural differences between student and teacher. By pulling the student's
RDMs towards those of the teacher, RDL significantly improved visual
classification performance when compared to baseline networks that did not use
transfer learning. In the future, RDL may enable combined supervised training
of deep neural networks using task constraints (e.g. images and category
labels) and constraints from brain-activity measurements, so as to build models
that replicate the internal representational spaces of biological brains.
|
Prediction of the Yield of Enzymatic Synthesis of Betulinic Acid Ester
Using Artificial Neural Networks and Support Vector Machine | 3\b{eta}-O-phthalic ester of betulinic acid is of great importance in
anticancer studies. However, the optimization of its reaction conditions
requires a large number of experimental works. To simplify the number of times
of optimization in experimental works, here, we use artificial neural network
(ANN) and support vector machine (SVM) models for the prediction of yields of
3\b{eta}-O-phthalic ester of betulinic acid synthesized by betulinic acid and
phthalic anhydride using lipase as biocatalyst. General regression neural
network (GRNN), multilayer feed-forward neural network (MLFN) and the SVM
models were trained based on experimental data. Four indicators were set as
independent variables, including time (h), temperature (C), amount of enzyme
(mg) and molar ratio, while the yield of the 3\b{eta}-O-phthalic ester of
betulinic acid was set as the dependent variable. Results show that the GRNN
and SVM models have the best prediction results during the testing process,
with comparatively low RMS errors (4.01 and 4.23respectively) and short
training times (both 1s). The prediction accuracy of the GRNN and SVM are both
100% in testing process, under the tolerance of 30%.
|
Block-diagonal covariance selection for high-dimensional Gaussian
graphical models | Gaussian graphical models are widely utilized to infer and visualize networks
of dependencies between continuous variables. However, inferring the graph is
difficult when the sample size is small compared to the number of variables. To
reduce the number of parameters to estimate in the model, we propose a
non-asymptotic model selection procedure supported by strong theoretical
guarantees based on an oracle inequality and a minimax lower bound. The
covariance matrix of the model is approximated by a block-diagonal matrix. The
structure of this matrix is detected by thresholding the sample covariance
matrix, where the threshold is selected using the slope heuristic. Based on the
block-diagonal structure of the covariance matrix, the estimation problem is
divided into several independent problems: subsequently, the network of
dependencies between variables is inferred using the graphical lasso algorithm
in each block. The performance of the procedure is illustrated on simulated
data. An application to a real gene expression dataset with a limited sample
size is also presented: the dimension reduction allows attention to be
objectively focused on interactions among smaller subsets of genes, leading to
a more parsimonious and interpretable modular network.
|
Kernel Methods for Accurate UWB-Based Ranging with Reduced Complexity | Accurate and robust positioning in multipath environments can enable many
applications, such as search-and-rescue and asset tracking. For this problem,
ultra-wideband (UWB) technology can provide the most accurate range estimates,
which are required for range-based positioning. However, UWB still faces a
problem with non-line-of-sight (NLOS) measurements, in which the range
estimates based on time-of-arrival (TOA) will typically be positively biased.
There are many techniques that address this problem, mainly based on NLOS
identification and NLOS error mitigation algorithms. However, these techniques
do not exploit all available information in the UWB channel impulse response.
Kernel-based machine learning methods, such as Gaussian Process Regression
(GPR), are able to make use of all information, but they may be too complex in
their original form. In this paper, we propose novel ranging methods based on
kernel principal component analysis (kPCA), in which the selected channel
parameters are projected onto a nonlinear orthogonal high-dimensional space,
and a subset of these projections is then used as an input for ranging. We
evaluate the proposed methods using real UWB measurements obtained in a
basement tunnel, and found that one of the proposed methods is able to
outperform state-of-the-art, even if little training samples are available.
|
Efficient non-greedy optimization of decision trees | Decision trees and randomized forests are widely used in computer vision and
machine learning. Standard algorithms for decision tree induction optimize the
split functions one node at a time according to some splitting criteria. This
greedy procedure often leads to suboptimal trees. In this paper, we present an
algorithm for optimizing the split functions at all levels of the tree jointly
with the leaf parameters, based on a global objective. We show that the problem
of finding optimal linear-combination (oblique) splits for decision trees is
related to structured prediction with latent variables, and we formulate a
convex-concave upper bound on the tree's empirical loss. The run-time of
computing the gradient of the proposed surrogate objective with respect to each
training exemplar is quadratic in the the tree depth, and thus training deep
trees is feasible. The use of stochastic gradient descent for optimization
enables effective training with large datasets. Experiments on several
classification benchmarks demonstrate that the resulting non-greedy decision
trees outperform greedy decision tree baselines.
|
Properly Learning Poisson Binomial Distributions in Almost Polynomial
Time | We give an algorithm for properly learning Poisson binomial distributions. A
Poisson binomial distribution (PBD) of order $n$ is the discrete probability
distribution of the sum of $n$ mutually independent Bernoulli random variables.
Given $\widetilde{O}(1/\epsilon^2)$ samples from an unknown PBD $\mathbf{p}$,
our algorithm runs in time $(1/\epsilon)^{O(\log \log (1/\epsilon))}$, and
outputs a hypothesis PBD that is $\epsilon$-close to $\mathbf{p}$ in total
variation distance. The previously best known running time for properly
learning PBDs was $(1/\epsilon)^{O(\log(1/\epsilon))}$.
As one of our main contributions, we provide a novel structural
characterization of PBDs. We prove that, for all $\epsilon >0,$ there exists an
explicit collection $\cal{M}$ of $(1/\epsilon)^{O(\log \log (1/\epsilon))}$
vectors of multiplicities, such that for any PBD $\mathbf{p}$ there exists a
PBD $\mathbf{q}$ with $O(\log(1/\epsilon))$ distinct parameters whose
multiplicities are given by some element of ${\cal M}$, such that $\mathbf{q}$
is $\epsilon$-close to $\mathbf{p}$. Our proof combines tools from Fourier
analysis and algebraic geometry.
Our approach to the proper learning problem is as follows: Starting with an
accurate non-proper hypothesis, we fit a PBD to this hypothesis. More
specifically, we essentially start with the hypothesis computed by the
computationally efficient non-proper learning algorithm in our recent
work~\cite{DKS15}. Our aforementioned structural characterization allows us to
reduce the corresponding fitting problem to a collection of
$(1/\epsilon)^{O(\log \log(1/\epsilon))}$ systems of low-degree polynomial
inequalities. We show that each such system can be solved in time
$(1/\epsilon)^{O(\log \log(1/\epsilon))}$, which yields the overall running
time of our algorithm.
|
LSTM-based Deep Learning Models for Non-factoid Answer Selection | In this paper, we apply a general deep learning (DL) framework for the answer
selection task, which does not depend on manually defined features or
linguistic tools. The basic framework is to build the embeddings of questions
and answers based on bidirectional long short-term memory (biLSTM) models, and
measure their closeness by cosine similarity. We further extend this basic
model in two directions. One direction is to define a more composite
representation for questions and answers by combining convolutional neural
network with the basic framework. The other direction is to utilize a simple
but efficient attention mechanism in order to generate the answer
representation according to the question context. Several variations of models
are provided. The models are examined by two datasets, including TREC-QA and
InsuranceQA. Experimental results demonstrate that the proposed models
substantially outperform several strong baselines.
|
Action Recognition using Visual Attention | We propose a soft attention based model for the task of action recognition in
videos. We use multi-layered Recurrent Neural Networks (RNNs) with Long
Short-Term Memory (LSTM) units which are deep both spatially and temporally.
Our model learns to focus selectively on parts of the video frames and
classifies videos after taking a few glimpses. The model essentially learns
which parts in the frames are relevant for the task at hand and attaches higher
importance to them. We evaluate the model on UCF-11 (YouTube Action), HMDB-51
and Hollywood2 datasets and analyze how the model focuses its attention
depending on the scene and the action being performed.
|
Seeing the Unseen Network: Inferring Hidden Social Ties from
Respondent-Driven Sampling | Learning about the social structure of hidden and hard-to-reach populations
--- such as drug users and sex workers --- is a major goal of epidemiological
and public health research on risk behaviors and disease prevention.
Respondent-driven sampling (RDS) is a peer-referral process widely used by many
health organizations, where research subjects recruit other subjects from their
social network. In such surveys, researchers observe who recruited whom, along
with the time of recruitment and the total number of acquaintances (network
degree) of respondents. However, due to privacy concerns, the identities of
acquaintances are not disclosed. In this work, we show how to reconstruct the
underlying network structure through which the subjects are recruited. We
formulate the dynamics of RDS as a continuous-time diffusion process over the
underlying graph and derive the likelihood for the recruitment time series
under an arbitrary recruitment time distribution. We develop an efficient
stochastic optimization algorithm called RENDER (REspoNdent-Driven nEtwork
Reconstruction) that finds the network that best explains the collected data.
We support our analytical results through an exhaustive set of experiments on
both synthetic and real data.
|
Deep Reinforcement Learning in Parameterized Action Space | Recent work has shown that deep neural networks are capable of approximating
both value functions and policies in reinforcement learning domains featuring
continuous state and action spaces. However, to the best of our knowledge no
previous work has succeeded at using deep neural networks in structured
(parameterized) continuous action spaces. To fill this gap, this paper focuses
on learning within the domain of simulated RoboCup soccer, which features a
small set of discrete action types, each of which is parameterized with
continuous variables. The best learned agent can score goals more reliably than
the 2012 RoboCup champion agent. As such, this paper represents a successful
extension of deep reinforcement learning to the class of parameterized action
space MDPs.
|
A Continuous-time Mutually-Exciting Point Process Framework for
Prioritizing Events in Social Media | The overwhelming amount and rate of information update in online social media
is making it increasingly difficult for users to allocate their attention to
their topics of interest, thus there is a strong need for prioritizing news
feeds. The attractiveness of a post to a user depends on many complex
contextual and temporal features of the post. For instance, the contents of the
post, the responsiveness of a third user, and the age of the post may all have
impact. So far, these static and dynamic features has not been incorporated in
a unified framework to tackle the post prioritization problem. In this paper,
we propose a novel approach for prioritizing posts based on a feature modulated
multi-dimensional point process. Our model is able to simultaneously capture
textual and sentiment features, and temporal features such as self-excitation,
mutual-excitation and bursty nature of social interaction. As an evaluation, we
also curated a real-world conversational benchmark dataset crawled from
Facebook. In our experiments, we demonstrate that our algorithm is able to
achieve the-state-of-the-art performance in terms of analyzing, predicting, and
prioritizing events. In terms of interpretability of our method, we observe
that features indicating individual user profile and linguistic characteristics
of the events work best for prediction and prioritization of new events.
|
Deep Mean Maps | The use of distributions and high-level features from deep architecture has
become commonplace in modern computer vision. Both of these methodologies have
separately achieved a great deal of success in many computer vision tasks.
However, there has been little work attempting to leverage the power of these
to methodologies jointly. To this end, this paper presents the Deep Mean Maps
(DMMs) framework, a novel family of methods to non-parametrically represent
distributions of features in convolutional neural network models.
DMMs are able to both classify images using the distribution of top-level
features, and to tune the top-level features for performing this task. We show
how to implement DMMs using a special mean map layer composed of typical CNN
operations, making both forward and backward propagation simple.
We illustrate the efficacy of DMMs at analyzing distributional patterns in
image data in a synthetic data experiment. We also show that we extending
existing deep architectures with DMMs improves the performance of existing CNNs
on several challenging real-world datasets.
|
Adaptive Affinity Matrix for Unsupervised Metric Learning | Spectral clustering is one of the most popular clustering approaches with the
capability to handle some challenging clustering problems. Most spectral
clustering methods provide a nonlinear map from the data manifold to a
subspace. Only a little work focuses on the explicit linear map which can be
viewed as the unsupervised distance metric learning. In practice, the selection
of the affinity matrix exhibits a tremendous impact on the unsupervised
learning. While much success of affinity learning has been achieved in recent
years, some issues such as noise reduction remain to be addressed. In this
paper, we propose a novel method, dubbed Adaptive Affinity Matrix (AdaAM), to
learn an adaptive affinity matrix and derive a distance metric from the
affinity. We assume the affinity matrix to be positive semidefinite with
ability to quantify the pairwise dissimilarity. Our method is based on posing
the optimization of objective function as a spectral decomposition problem. We
yield the affinity from both the original data distribution and the widely-used
heat kernel. The provided matrix can be regarded as the optimal representation
of pairwise relationship on the manifold. Extensive experiments on a number of
real-world data sets show the effectiveness and efficiency of AdaAM.
|
Neuroprosthetic decoder training as imitation learning | Neuroprosthetic brain-computer interfaces function via an algorithm which
decodes neural activity of the user into movements of an end effector, such as
a cursor or robotic arm. In practice, the decoder is often learned by updating
its parameters while the user performs a task. When the user's intention is not
directly observable, recent methods have demonstrated value in training the
decoder against a surrogate for the user's intended movement. We describe how
training a decoder in this way is a novel variant of an imitation learning
problem, where an oracle or expert is employed for supervised training in lieu
of direct observations, which are not available. Specifically, we describe how
a generic imitation learning meta-algorithm, dataset aggregation (DAgger, [1]),
can be adapted to train a generic brain-computer interface. By deriving
existing learning algorithms for brain-computer interfaces in this framework,
we provide a novel analysis of regret (an important metric of learning
efficacy) for brain-computer interfaces. This analysis allows us to
characterize the space of algorithmic variants and bounds on their regret
rates. Existing approaches for decoder learning have been performed in the
cursor control setting, but the available design principles for these decoders
are such that it has been impossible to scale them to naturalistic settings.
Leveraging our findings, we then offer an algorithm that combines imitation
learning with optimal control, which should allow for training of arbitrary
effectors for which optimal control can generate goal-oriented control. We
demonstrate this novel and general BCI algorithm with simulated neuroprosthetic
control of a 26 degree-of-freedom model of an arm, a sophisticated and
realistic end effector.
|
On the Quality of the Initial Basin in Overspecified Neural Networks | Deep learning, in the form of artificial neural networks, has achieved
remarkable practical success in recent years, for a variety of difficult
machine learning applications. However, a theoretical explanation for this
remains a major open problem, since training neural networks involves
optimizing a highly non-convex objective function, and is known to be
computationally hard in the worst case. In this work, we study the
\emph{geometric} structure of the associated non-convex objective function, in
the context of ReLU networks and starting from a random initialization of the
network parameters. We identify some conditions under which it becomes more
favorable to optimization, in the sense of (i) High probability of initializing
at a point from which there is a monotonically decreasing path to a global
minimum; and (ii) High probability of initializing at a basin (suitably
defined) with a small minimal objective value. A common theme in our results is
that such properties are more likely to hold for larger ("overspecified")
networks, which accords with some recent empirical and theoretical
observations.
|
Active Contextual Entropy Search | Contextual policy search allows adapting robotic movement primitives to
different situations. For instance, a locomotion primitive might be adapted to
different terrain inclinations or desired walking speeds. Such an adaptation is
often achievable by modifying a small number of hyperparameters. However,
learning, when performed on real robotic systems, is typically restricted to a
small number of trials. Bayesian optimization has recently been proposed as a
sample-efficient means for contextual policy search that is well suited under
these conditions. In this work, we extend entropy search, a variant of Bayesian
optimization, such that it can be used for active contextual policy search
where the agent selects those tasks during training in which it expects to
learn the most. Empirical results in simulation suggest that this allows
learning successful behavior with less trials.
|
Deep Feature Learning for EEG Recordings | We introduce and compare several strategies for learning discriminative
features from electroencephalography (EEG) recordings using deep learning
techniques. EEG data are generally only available in small quantities, they are
high-dimensional with a poor signal-to-noise ratio, and there is considerable
variability between individual subjects and recording sessions. Our proposed
techniques specifically address these challenges for feature learning.
Cross-trial encoding forces auto-encoders to focus on features that are stable
across trials. Similarity-constraint encoders learn features that allow to
distinguish between classes by demanding that two trials from the same class
are more similar to each other than to trials from other classes. This
tuple-based training approach is especially suitable for small datasets.
Hydra-nets allow for separate processing pathways adapting to subsets of a
dataset and thus combine the advantages of individual feature learning (better
adaptation of early, low-level processing) with group model training (better
generalization of higher-level processing in deeper layers). This way, models
can, for instance, adapt to each subject individually to compensate for
differences in spatial patterns due to anatomical differences or variance in
electrode positions. The different techniques are evaluated using the publicly
available OpenMIIR dataset of EEG recordings taken while participants listened
to and imagined music.
|
Handling Class Imbalance in Link Prediction using Learning to Rank
Techniques | We consider the link prediction problem in a partially observed network,
where the objective is to make predictions in the unobserved portion of the
network. Many existing methods reduce link prediction to binary classification
problem. However, the dominance of absent links in real world networks makes
misclassification error a poor performance metric. Instead, researchers have
argued for using ranking performance measures, like AUC, AP and NDCG, for
evaluation. Our main contribution is to recast the link prediction problem as a
learning to rank problem and use effective learning to rank techniques directly
during training. This is in contrast to existing work that uses ranking
measures only during evaluation. Our approach is able to deal with the class
imbalance problem by using effective, scalable learning to rank techniques
during training. Furthermore, our approach allows us to combine network
topology and node features. As a demonstration of our general approach, we
develop a link prediction method by optimizing the cross-entropy surrogate,
originally used in the popular ListNet ranking algorithm. We conduct extensive
experiments on publicly available co-authorship, citation and metabolic
networks to demonstrate the merits of our method.
|
Similarity-based Text Recognition by Deeply Supervised Siamese Network | In this paper, we propose a new text recognition model based on measuring the
visual similarity of text and predicting the content of unlabeled texts. First
a Siamese convolutional network is trained with deep supervision on a labeled
training dataset. This network projects texts into a similarity manifold. The
Deeply Supervised Siamese network learns visual similarity of texts. Then a
K-nearest neighbor classifier is used to predict unlabeled text based on
similarity distance to labeled texts. The performance of the model is evaluated
on three datasets of machine-print and hand-written text combined. We
demonstrate that the model reduces the cost of human estimation by $50\%-85\%$.
The error of the system is less than $0.5\%$. The proposed model outperform
conventional Siamese network by finding visually-similar barely-readable and
readable text, e.g. machine-printed, handwritten, due to deep supervision. The
results also demonstrate that the predicted labels are sometimes better than
human labels e.g. spelling correction.
|
Symbol Grounding Association in Multimodal Sequences with Missing
Elements | In this paper, we extend a symbolic association framework for being able to
handle missing elements in multimodal sequences. The general scope of the work
is the symbolic associations of object-word mappings as it happens in language
development in infants. In other words, two different representations of the
same abstract concepts can associate in both directions. This scenario has been
long interested in Artificial Intelligence, Psychology, and Neuroscience. In
this work, we extend a recent approach for multimodal sequences (visual and
audio) to also cope with missing elements in one or both modalities. Our method
uses two parallel Long Short-Term Memories (LSTMs) with a learning rule based
on EM-algorithm. It aligns both LSTM outputs via Dynamic Time Warping (DTW). We
propose to include an extra step for the combination with the max operation for
exploiting the common elements between both sequences. The motivation behind is
that the combination acts as a condition selector for choosing the best
representation from both LSTMs. We evaluated the proposed extension in the
following scenarios: missing elements in one modality (visual or audio) and
missing elements in both modalities (visual and sound). The performance of our
extension reaches better results than the original model and similar results to
individual LSTM trained in each modality.
|
Dynamic Sum Product Networks for Tractable Inference on Sequence Data
(Extended Version) | Sum-Product Networks (SPN) have recently emerged as a new class of tractable
probabilistic graphical models. Unlike Bayesian networks and Markov networks
where inference may be exponential in the size of the network, inference in
SPNs is in time linear in the size of the network. Since SPNs represent
distributions over a fixed set of variables only, we propose dynamic sum
product networks (DSPNs) as a generalization of SPNs for sequence data of
varying length. A DSPN consists of a template network that is repeated as many
times as needed to model data sequences of any length. We present a local
search technique to learn the structure of the template network. In contrast to
dynamic Bayesian networks for which inference is generally exponential in the
number of variables per time slice, DSPNs inherit the linear inference
complexity of SPNs. We demonstrate the advantages of DSPNs over DBNs and other
models on several datasets of sequence data.
|
Deeply-Recursive Convolutional Network for Image Super-Resolution | We propose an image super-resolution method (SR) using a deeply-recursive
convolutional network (DRCN). Our network has a very deep recursive layer (up
to 16 recursions). Increasing recursion depth can improve performance without
introducing new parameters for additional convolutions. Albeit advantages,
learning a DRCN is very hard with a standard gradient descent method due to
exploding/vanishing gradients. To ease the difficulty of training, we propose
two extensions: recursive-supervision and skip-connection. Our method
outperforms previous methods by a large margin.
|
Distillation as a Defense to Adversarial Perturbations against Deep
Neural Networks | Deep learning algorithms have been shown to perform extremely well on many
classical machine learning problems. However, recent studies have shown that
deep learning, like other machine learning techniques, is vulnerable to
adversarial samples: inputs crafted to force a deep neural network (DNN) to
provide adversary-selected outputs. Such attacks can seriously undermine the
security of the system supported by the DNN, sometimes with devastating
consequences. For example, autonomous vehicles can be crashed, illicit or
illegal content can bypass content filters, or biometric authentication systems
can be manipulated to allow improper access. In this work, we introduce a
defensive mechanism called defensive distillation to reduce the effectiveness
of adversarial samples on DNNs. We analytically investigate the
generalizability and robustness properties granted by the use of defensive
distillation when training DNNs. We also empirically study the effectiveness of
our defense mechanisms on two DNNs placed in adversarial settings. The study
shows that defensive distillation can reduce effectiveness of sample creation
from 95% to less than 0.5% on a studied DNN. Such dramatic gains can be
explained by the fact that distillation leads gradients used in adversarial
sample creation to be reduced by a factor of 10^30. We also find that
distillation increases the average minimum number of features that need to be
modified to create adversarial samples by about 800% on one of the DNNs we
tested.
|
Sparse Nonlinear Regression: Parameter Estimation and Asymptotic
Inference | We study parameter estimation and asymptotic inference for sparse nonlinear
regression. More specifically, we assume the data are given by $y = f( x^\top
\beta^* ) + \epsilon$, where $f$ is nonlinear. To recover $\beta^*$, we propose
an $\ell_1$-regularized least-squares estimator. Unlike classical linear
regression, the corresponding optimization problem is nonconvex because of the
nonlinearity of $f$. In spite of the nonconvexity, we prove that under mild
conditions, every stationary point of the objective enjoys an optimal
statistical rate of convergence. In addition, we provide an efficient algorithm
that provably converges to a stationary point. We also access the uncertainty
of the obtained estimator. Specifically, based on any stationary point of the
objective, we construct valid hypothesis tests and confidence intervals for the
low dimensional components of the high-dimensional parameter $\beta^*$.
Detailed numerical results are provided to back up our theory.
|
Efficient Training of Very Deep Neural Networks for Supervised Hashing | In this paper, we propose training very deep neural networks (DNNs) for
supervised learning of hash codes. Existing methods in this context train
relatively "shallow" networks limited by the issues arising in back propagation
(e.e. vanishing gradients) as well as computational efficiency. We propose a
novel and efficient training algorithm inspired by alternating direction method
of multipliers (ADMM) that overcomes some of these limitations. Our method
decomposes the training process into independent layer-wise local updates
through auxiliary variables. Empirically we observe that our training algorithm
always converges and its computational complexity is linearly proportional to
the number of edges in the networks. Empirically we manage to train DNNs with
64 hidden layers and 1024 nodes per layer for supervised hashing in about 3
hours using a single GPU. Our proposed very deep supervised hashing (VDSH)
method significantly outperforms the state-of-the-art on several benchmark
datasets.
|
8-Bit Approximations for Parallelism in Deep Learning | The creation of practical deep learning data-products often requires
parallelization across processors and computers to make deep learning feasible
on large data sets, but bottlenecks in communication bandwidth make it
difficult to attain good speedups through parallelism. Here we develop and test
8-bit approximation algorithms which make better use of the available bandwidth
by compressing 32-bit gradients and nonlinear activations to 8-bit
approximations. We show that these approximations do not decrease predictive
performance on MNIST, CIFAR10, and ImageNet for both model and data parallelism
and provide a data transfer speedup of 2x relative to 32-bit parallelism. We
build a predictive model for speedups based on our experimental data, verify
its validity on known speedup data, and show that we can obtain a speedup of
50x and more on a system of 96 GPUs compared to a speedup of 23x for 32-bit. We
compare our data types with other methods and show that 8-bit approximations
achieve state-of-the-art speedups for model parallelism. Thus 8-bit
approximation is an efficient method to parallelize convolutional networks on
very large systems of GPUs.
|
A Test of Relative Similarity For Model Selection in Generative Models | Probabilistic generative models provide a powerful framework for representing
data that avoids the expense of manual annotation typically needed by
discriminative approaches. Model selection in this generative setting can be
challenging, however, particularly when likelihoods are not easily accessible.
To address this issue, we introduce a statistical test of relative similarity,
which is used to determine which of two models generates samples that are
significantly closer to a real-world reference dataset of interest. We use as
our test statistic the difference in maximum mean discrepancies (MMDs) between
the reference dataset and each model dataset, and derive a powerful,
low-variance test based on the joint asymptotic distribution of the MMDs
between each reference-model pair. In experiments on deep generative models,
including the variational auto-encoder and generative moment matching network,
the tests provide a meaningful ranking of model performance as a function of
parameter and training settings.
|
Accurate Image Super-Resolution Using Very Deep Convolutional Networks | We present a highly accurate single-image super-resolution (SR) method. Our
method uses a very deep convolutional network inspired by VGG-net used for
ImageNet classification \cite{simonyan2015very}. We find increasing our network
depth shows a significant improvement in accuracy. Our final model uses 20
weight layers. By cascading small filters many times in a deep network
structure, contextual information over large image regions is exploited in an
efficient way. With very deep networks, however, convergence speed becomes a
critical issue during training. We propose a simple yet effective training
procedure. We learn residuals only and use extremely high learning rates
($10^4$ times higher than SRCNN \cite{dong2015image}) enabled by adjustable
gradient clipping. Our proposed method performs better than existing methods in
accuracy and visual improvements in our results are easily noticeable.
|
DeepFool: a simple and accurate method to fool deep neural networks | State-of-the-art deep neural networks have achieved impressive results on
many image classification tasks. However, these same architectures have been
shown to be unstable to small, well sought, perturbations of the images.
Despite the importance of this phenomenon, no effective methods have been
proposed to accurately compute the robustness of state-of-the-art deep
classifiers to such perturbations on large-scale datasets. In this paper, we
fill this gap and propose the DeepFool algorithm to efficiently compute
perturbations that fool deep networks, and thus reliably quantify the
robustness of these classifiers. Extensive experimental results show that our
approach outperforms recent methods in the task of computing adversarial
perturbations and making classifiers more robust.
|
Deep Reinforcement Learning with a Natural Language Action Space | This paper introduces a novel architecture for reinforcement learning with
deep neural networks designed to handle state and action spaces characterized
by natural language, as found in text-based games. Termed a deep reinforcement
relevance network (DRRN), the architecture represents action and state spaces
with separate embedding vectors, which are combined with an interaction
function to approximate the Q-function in reinforcement learning. We evaluate
the DRRN on two popular text games, showing superior performance over other
deep Q-learning architectures. Experiments with paraphrased action descriptions
show that the model is extracting meaning rather than simply memorizing strings
of text.
|
Deep Activity Recognition Models with Triaxial Accelerometers | Despite the widespread installation of accelerometers in almost all mobile
phones and wearable devices, activity recognition using accelerometers is still
immature due to the poor recognition accuracy of existing recognition methods
and the scarcity of labeled training data. We consider the problem of human
activity recognition using triaxial accelerometers and deep learning paradigms.
This paper shows that deep activity recognition models (a) provide better
recognition accuracy of human activities, (b) avoid the expensive design of
handcrafted features in existing systems, and (c) utilize the massive unlabeled
acceleration samples for unsupervised feature extraction. Moreover, a hybrid
approach of deep learning and hidden Markov models (DL-HMM) is presented for
sequential activity recognition. This hybrid approach integrates the
hierarchical representations of deep activity recognition models with the
stochastic modeling of temporal sequences in the hidden Markov models. We show
substantial recognition improvement on real world datasets over
state-of-the-art methods of human activity recognition using triaxial
accelerometers.
|
Robust Elastic Net Regression | We propose a robust elastic net (REN) model for high-dimensional sparse
regression and give its performance guarantees (both the statistical error
bound and the optimization bound). A simple idea of trimming the inner product
is applied to the elastic net model. Specifically, we robustify the covariance
matrix by trimming the inner product based on the intuition that the trimmed
inner product can not be significant affected by a bounded number of
arbitrarily corrupted points (outliers). The REN model can also derive two
interesting special cases: robust Lasso and robust soft thresholding.
Comprehensive experimental results show that the robustness of the proposed
model consistently outperforms the original elastic net and matches the
performance guarantees nicely.
|
An Iterative Reweighted Method for Tucker Decomposition of Incomplete
Multiway Tensors | We consider the problem of low-rank decomposition of incomplete multiway
tensors. Since many real-world data lie on an intrinsically low dimensional
subspace, tensor low-rank decomposition with missing entries has applications
in many data analysis problems such as recommender systems and image
inpainting. In this paper, we focus on Tucker decomposition which represents an
Nth-order tensor in terms of N factor matrices and a core tensor via
multilinear operations. To exploit the underlying multilinear low-rank
structure in high-dimensional datasets, we propose a group-based log-sum
penalty functional to place structural sparsity over the core tensor, which
leads to a compact representation with smallest core tensor. The method for
Tucker decomposition is developed by iteratively minimizing a surrogate
function that majorizes the original objective function, which results in an
iterative reweighted process. In addition, to reduce the computational
complexity, an over-relaxed monotone fast iterative shrinkage-thresholding
technique is adapted and embedded in the iterative reweighted process. The
proposed method is able to determine the model complexity (i.e. multilinear
rank) in an automatic way. Simulation results show that the proposed algorithm
offers competitive performance compared with other existing algorithms.
|
Deep Linear Discriminant Analysis | We introduce Deep Linear Discriminant Analysis (DeepLDA) which learns
linearly separable latent representations in an end-to-end fashion. Classic LDA
extracts features which preserve class separability and is used for
dimensionality reduction for many classification problems. The central idea of
this paper is to put LDA on top of a deep neural network. This can be seen as a
non-linear extension of classic LDA. Instead of maximizing the likelihood of
target labels for individual samples, we propose an objective function that
pushes the network to produce feature distributions which: (a) have low
variance within the same class and (b) high variance between different classes.
Our objective is derived from the general LDA eigenvalue problem and still
allows to train with stochastic gradient descent and back-propagation. For
evaluation we test our approach on three different benchmark datasets (MNIST,
CIFAR-10 and STL-10). DeepLDA produces competitive results on MNIST and
CIFAR-10 and outperforms a network trained with categorical cross entropy (same
architecture) on a supervised setting of STL-10.
|
Learning Representations of Affect from Speech | There has been a lot of prior work on representation learning for speech
recognition applications, but not much emphasis has been given to an
investigation of effective representations of affect from speech, where the
paralinguistic elements of speech are separated out from the verbal content. In
this paper, we explore denoising autoencoders for learning paralinguistic
attributes i.e. categorical and dimensional affective traits from speech. We
show that the representations learnt by the bottleneck layer of the autoencoder
are highly discriminative of activation intensity and at separating out
negative valence (sadness and anger) from positive valence (happiness). We
experiment with different input speech features (such as FFT and log-mel
spectrograms with temporal context windows), and different autoencoder
architectures (such as stacked and deep autoencoders). We also learn utterance
specific representations by a combination of denoising autoencoders and BLSTM
based recurrent autoencoders. Emotion classification is performed with the
learnt temporal/dynamic representations to evaluate the quality of the
representations. Experiments on a well-established real-life speech dataset
(IEMOCAP) show that the learnt representations are comparable to state of the
art feature extractors (such as voice quality features and MFCCs) and are
competitive with state-of-the-art approaches at emotion and dimensional affect
recognition.
|
Large-Scale Approximate Kernel Canonical Correlation Analysis | Kernel canonical correlation analysis (KCCA) is a nonlinear multi-view
representation learning technique with broad applicability in statistics and
machine learning. Although there is a closed-form solution for the KCCA
objective, it involves solving an $N\times N$ eigenvalue system where $N$ is
the training set size, making its computational requirements in both memory and
time prohibitive for large-scale problems. Various approximation techniques
have been developed for KCCA. A commonly used approach is to first transform
the original inputs to an $M$-dimensional random feature space so that inner
products in the feature space approximate kernel evaluations, and then apply
linear CCA to the transformed inputs. In many applications, however, the
dimensionality $M$ of the random feature space may need to be very large in
order to obtain a sufficiently good approximation; it then becomes challenging
to perform the linear CCA step on the resulting very high-dimensional data
matrices. We show how to use a stochastic optimization algorithm, recently
proposed for linear CCA and its neural-network extension, to further alleviate
the computation requirements of approximate KCCA. This approach allows us to
run approximate KCCA on a speech dataset with $1.4$ million training samples
and a random feature space of dimensionality $M=100000$ on a typical
workstation.
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.