title
stringlengths 7
246
| abstract
stringlengths 6
3.31k
|
|---|---|
Expressive recommender systems through normalized nonnegative models | We introduce normalized nonnegative models (NNM) for explorative data
analysis. NNMs are partial convexifications of models from probability theory.
We demonstrate their value at the example of item recommendation. We show that
NNM-based recommender systems satisfy three criteria that all recommender
systems should ideally satisfy: high predictive power, computational
tractability, and expressive representations of users and items. Expressive
user and item representations are important in practice to succinctly summarize
the pool of customers and the pool of items. In NNMs, user representations are
expressive because each user's preference can be regarded as normalized mixture
of preferences of stereotypical users. The interpretability of item and user
representations allow us to arrange properties of items (e.g., genres of movies
or topics of documents) or users (e.g., personality traits) hierarchically.
|
Mixtures of Sparse Autoregressive Networks | We consider high-dimensional distribution estimation through autoregressive
networks. By combining the concepts of sparsity, mixtures and parameter sharing
we obtain a simple model which is fast to train and which achieves
state-of-the-art or better results on several standard benchmark datasets.
Specifically, we use an L1-penalty to regularize the conditional distributions
and introduce a procedure for automatic parameter sharing between mixture
components. Moreover, we propose a simple distributed representation which
permits exact likelihood evaluations since the latent variables are interleaved
with the observable variables and can be easily integrated out. Our model
achieves excellent generalization performance and scales well to extremely high
dimensions.
|
Causal interpretation rules for encoding and decoding models in
neuroimaging | Causal terminology is often introduced in the interpretation of encoding and
decoding models trained on neuroimaging data. In this article, we investigate
which causal statements are warranted and which ones are not supported by
empirical evidence. We argue that the distinction between encoding and decoding
models is not sufficient for this purpose: relevant features in encoding and
decoding models carry a different meaning in stimulus- and in response-based
experimental paradigms. We show that only encoding models in the stimulus-based
setting support unambiguous causal interpretations. By combining encoding and
decoding models trained on the same data, however, we obtain insights into
causal relations beyond those that are implied by each individual model type.
We illustrate the empirical relevance of our theoretical findings on EEG data
recorded during a visuo-motor learning task.
|
Budget Online Multiple Kernel Learning | Online learning with multiple kernels has gained increasing interests in
recent years and found many applications. For classification tasks, Online
Multiple Kernel Classification (OMKC), which learns a kernel based classifier
by seeking the optimal linear combination of a pool of single kernel
classifiers in an online fashion, achieves superior accuracy and enjoys great
flexibility compared with traditional single-kernel classifiers. Despite being
studied extensively, existing OMKC algorithms suffer from high computational
cost due to their unbounded numbers of support vectors. To overcome this
drawback, we present a novel framework of Budget Online Multiple Kernel
Learning (BOMKL) and propose a new Sparse Passive Aggressive learning to
perform effective budget online learning. Specifically, we adopt a simple yet
effective Bernoulli sampling to decide if an incoming instance should be added
to the current set of support vectors. By limiting the number of support
vectors, our method can significantly accelerate OMKC while maintaining
satisfactory accuracy that is comparable to that of the existing OMKC
algorithms. We theoretically prove that our new method achieves an optimal
regret bound in expectation, and empirically found that the proposed algorithm
outperforms various OMKC algorithms and can easily scale up to large-scale
datasets.
|
Neural Programmer: Inducing Latent Programs with Gradient Descent | Deep neural networks have achieved impressive supervised classification
performance in many tasks including image recognition, speech recognition, and
sequence to sequence learning. However, this success has not been translated to
applications like question answering that may involve complex arithmetic and
logic reasoning. A major limitation of these models is in their inability to
learn even simple arithmetic and logic operations. For example, it has been
shown that neural networks fail to learn to add two binary numbers reliably. In
this work, we propose Neural Programmer, an end-to-end differentiable neural
network augmented with a small set of basic arithmetic and logic operations.
Neural Programmer can call these augmented operations over several steps,
thereby inducing compositional programs that are more complex than the built-in
operations. The model learns from a weak supervision signal which is the result
of execution of the correct program, hence it does not require expensive
annotation of the correct program itself. The decisions of what operations to
call, and what data segments to apply to are inferred by Neural Programmer.
Such decisions, during training, are done in a differentiable fashion so that
the entire network can be trained jointly by gradient descent. We find that
training the model is difficult, but it can be greatly improved by adding
random noise to the gradient. On a fairly complex synthetic table-comprehension
dataset, traditional recurrent networks and attentional models perform poorly
while Neural Programmer typically obtains nearly perfect accuracy.
|
Nonparametric Canonical Correlation Analysis | Canonical correlation analysis (CCA) is a classical representation learning
technique for finding correlated variables in multi-view data. Several
nonlinear extensions of the original linear CCA have been proposed, including
kernel and deep neural network methods. These approaches seek maximally
correlated projections among families of functions, which the user specifies
(by choosing a kernel or neural network structure), and are computationally
demanding. Interestingly, the theory of nonlinear CCA, without functional
restrictions, had been studied in the population setting by Lancaster already
in the 1950s, but these results have not inspired practical algorithms. We
revisit Lancaster's theory to devise a practical algorithm for nonparametric
CCA (NCCA). Specifically, we show that the solution can be expressed in terms
of the singular value decomposition of a certain operator associated with the
joint density of the views. Thus, by estimating the population density from
data, NCCA reduces to solving an eigenvalue system, superficially like kernel
CCA but, importantly, without requiring the inversion of any kernel matrix. We
also derive a partially linear CCA (PLCCA) variant in which one of the views
undergoes a linear projection while the other is nonparametric. Using a kernel
density estimate based on a small number of nearest neighbors, our NCCA and
PLCCA algorithms are memory-efficient, often run much faster, and perform
better than kernel CCA and comparable to deep CCA.
|
Deep learning is a good steganalysis tool when embedding key is reused
for different images, even if there is a cover source-mismatch | Since the BOSS competition, in 2010, most steganalysis approaches use a
learning methodology involving two steps: feature extraction, such as the Rich
Models (RM), for the image representation, and use of the Ensemble Classifier
(EC) for the learning step. In 2015, Qian et al. have shown that the use of a
deep learning approach that jointly learns and computes the features, is very
promising for the steganalysis. In this paper, we follow-up the study of Qian
et al., and show that, due to intrinsic joint minimization, the results
obtained from a Convolutional Neural Network (CNN) or a Fully Connected Neural
Network (FNN), if well parameterized, surpass the conventional use of a RM with
an EC. First, numerous experiments were conducted in order to find the best "
shape " of the CNN. Second, experiments were carried out in the clairvoyant
scenario in order to compare the CNN and FNN to an RM with an EC. The results
show more than 16% reduction in the classification error with our CNN or FNN.
Third, experiments were also performed in a cover-source mismatch setting. The
results show that the CNN and FNN are naturally robust to the mismatch problem.
In Addition to the experiments, we provide discussions on the internal
mechanisms of a CNN, and weave links with some previously stated ideas, in
order to understand the impressive results we obtained.
|
A Neural Transducer | Sequence-to-sequence models have achieved impressive results on various
tasks. However, they are unsuitable for tasks that require incremental
predictions to be made as more data arrives or tasks that have long input
sequences and output sequences. This is because they generate an output
sequence conditioned on an entire input sequence. In this paper, we present a
Neural Transducer that can make incremental predictions as more input arrives,
without redoing the entire computation. Unlike sequence-to-sequence models, the
Neural Transducer computes the next-step distribution conditioned on the
partially observed input sequence and the partially generated sequence. At each
time step, the transducer can decide to emit zero to many output symbols. The
data can be processed using an encoder and presented as input to the
transducer. The discrete decision to emit a symbol at every time step makes it
difficult to learn with conventional backpropagation. It is however possible to
train the transducer by using a dynamic programming algorithm to generate
target discrete decisions. Our experiments show that the Neural Transducer
works well in settings where it is required to produce output predictions as
data come in. We also find that the Neural Transducer performs well for long
sequences even when attention mechanisms are not used.
|
Sherlock: Scalable Fact Learning in Images | We study scalable and uniform understanding of facts in images. Existing
visual recognition systems are typically modeled differently for each fact type
such as objects, actions, and interactions. We propose a setting where all
these facts can be modeled simultaneously with a capacity to understand
unbounded number of facts in a structured way. The training data comes as
structured facts in images, including (1) objects (e.g., $<$boy$>$), (2)
attributes (e.g., $<$boy, tall$>$), (3) actions (e.g., $<$boy, playing$>$), and
(4) interactions (e.g., $<$boy, riding, a horse $>$). Each fact has a semantic
language view (e.g., $<$ boy, playing$>$) and a visual view (an image with this
fact). We show that learning visual facts in a structured way enables not only
a uniform but also generalizable visual understanding. We propose and
investigate recent and strong approaches from the multiview learning literature
and also introduce two learning representation models as potential baselines.
We applied the investigated methods on several datasets that we augmented with
structured facts and a large scale dataset of more than 202,000 facts and
814,000 images. Our experiments show the advantage of relating facts by the
structure by the proposed models compared to the designed baselines on
bidirectional fact retrieval.
|
Performing Highly Accurate Predictions Through Convolutional Networks
for Actual Telecommunication Challenges | We investigated how the application of deep learning, specifically the use of
convolutional networks trained with GPUs, can help to build better predictive
models in telecommunication business environments, and fill this gap. In
particular, we focus on the non-trivial problem of predicting customer churn in
telecommunication operators. Our model, called WiseNet, consists of a
convolutional network and a novel encoding method that transforms customer
activity data and Call Detail Records (CDRs) into images. Experimental
evaluation with several machine learning classifiers supports the ability of
WiseNet for learning features when using structured input data. For this type
of telecommunication business problems, we found that WiseNet outperforms
machine learning models with hand-crafted features, and does not require the
labor-intensive step of feature engineering. Furthermore, the same model has
been applied without retraining to a different market, achieving consistent
results. This confirms the generalization property of WiseNet and the ability
to extract useful representations.
|
A genetic algorithm to discover flexible motifs with support | Finding repeated patterns or motifs in a time series is an important
unsupervised task that has still a number of open issues, starting by the
definition of motif. In this paper, we revise the notion of motif support,
characterizing it as the number of patterns or repetitions that define a motif.
We then propose GENMOTIF, a genetic algorithm to discover motifs with support
which, at the same time, is flexible enough to accommodate other motif
specifications and task characteristics. GENMOTIF is an anytime algorithm that
easily adapts to many situations: searching in a range of segment lengths,
applying uniform scaling, dealing with multiple dimensions, using different
similarity and grouping criteria, etc. GENMOTIF is also parameter-friendly: it
has only two intuitive parameters which, if set within reasonable bounds, do
not substantially affect its performance. We demonstrate the value of our
approach in a number of synthetic and real-world settings, considering traffic
volume measurements, accelerometer signals, and telephone call records.
|
An Exploration of Softmax Alternatives Belonging to the Spherical Loss
Family | In a multi-class classification problem, it is standard to model the output
of a neural network as a categorical distribution conditioned on the inputs.
The output must therefore be positive and sum to one, which is traditionally
enforced by a softmax. This probabilistic mapping allows to use the maximum
likelihood principle, which leads to the well-known log-softmax loss. However
the choice of the softmax function seems somehow arbitrary as there are many
other possible normalizing functions. It is thus unclear why the log-softmax
loss would perform better than other loss alternatives. In particular Vincent
et al. (2015) recently introduced a class of loss functions, called the
spherical family, for which there exists an efficient algorithm to compute the
updates of the output weights irrespective of the output size. In this paper,
we explore several loss functions from this family as possible alternatives to
the traditional log-softmax. In particular, we focus our investigation on
spherical bounds of the log-softmax loss and on two spherical log-likelihood
losses, namely the log-Spherical Softmax suggested by Vincent et al. (2015) and
the log-Taylor Softmax that we introduce. Although these alternatives do not
yield as good results as the log-softmax loss on two language modeling tasks,
they surprisingly outperform it in our experiments on MNIST and CIFAR-10,
suggesting that they might be relevant in a broad range of applications.
|
Diversity Networks: Neural Network Compression Using Determinantal Point
Processes | We introduce Divnet, a flexible technique for learning networks with diverse
neurons. Divnet models neuronal diversity by placing a Determinantal Point
Process (DPP) over neurons in a given layer. It uses this DPP to select a
subset of diverse neurons and subsequently fuses the redundant neurons into the
selected ones. Compared with previous approaches, Divnet offers a more
principled, flexible technique for capturing neuronal diversity and thus
implicitly enforcing regularization. This enables effective auto-tuning of
network architecture and leads to smaller network sizes without hurting
performance. Moreover, through its focus on diversity and neuron fusing, Divnet
remains compatible with other procedures that seek to reduce memory footprints
of networks. We present experimental results to corroborate our claims: for
pruning neural networks, Divnet is seen to be notably superior to competing
approaches.
|
Topic Modeling of Behavioral Modes Using Sensor Data | The field of Movement Ecology, like so many other fields, is experiencing a
period of rapid growth in availability of data. As the volume rises,
traditional methods are giving way to machine learning and data science, which
are playing an increasingly large part it turning this data into
science-driving insights. One rich and interesting source is the bio-logger.
These small electronic wearable devices are attached to animals free to roam in
their natural habitats, and report back readings from multiple sensors,
including GPS and accelerometer bursts. A common use of accelerometer data is
for supervised learning of behavioral modes. However, we need unsupervised
analysis tools as well, in order to overcome the inherent difficulties of
obtaining a labeled dataset, which in some cases is either infeasible or does
not successfully encompass the full repertoire of behavioral modes of interest.
Here we present a matrix factorization based topic-model method for
accelerometer bursts, derived using a linear mixture property of patch
features. Our method is validated via comparison to a labeled dataset, and is
further compared to standard clustering algorithms.
|
Yin and Yang: Balancing and Answering Binary Visual Questions | The complex compositional structure of language makes problems at the
intersection of vision and language challenging. But language also provides a
strong prior that can result in good superficial performance, without the
underlying models truly understanding the visual content. This can hinder
progress in pushing state of art in the computer vision aspects of multi-modal
AI. In this paper, we address binary Visual Question Answering (VQA) on
abstract scenes. We formulate this problem as visual verification of concepts
inquired in the questions. Specifically, we convert the question to a tuple
that concisely summarizes the visual concept to be detected in the image. If
the concept can be found in the image, the answer to the question is "yes", and
otherwise "no". Abstract scenes play two roles (1) They allow us to focus on
the high-level semantics of the VQA task as opposed to the low-level
recognition problems, and perhaps more importantly, (2) They provide us the
modality to balance the dataset such that language priors are controlled, and
the role of vision is essential. In particular, we collect fine-grained pairs
of scenes for every question, such that the answer to the question is "yes" for
one scene, and "no" for the other for the exact same question. Indeed, language
priors alone do not perform better than chance on our balanced dataset.
Moreover, our proposed approach matches the performance of a state-of-the-art
VQA approach on the unbalanced dataset, and outperforms it on the balanced
dataset.
|
How (not) to Train your Generative Model: Scheduled Sampling,
Likelihood, Adversary? | Modern applications and progress in deep learning research have created
renewed interest for generative models of text and of images. However, even
today it is unclear what objective functions one should use to train and
evaluate these models. In this paper we present two contributions.
Firstly, we present a critique of scheduled sampling, a state-of-the-art
training method that contributed to the winning entry to the MSCOCO image
captioning benchmark in 2015. Here we show that despite this impressive
empirical performance, the objective function underlying scheduled sampling is
improper and leads to an inconsistent learning algorithm.
Secondly, we revisit the problems that scheduled sampling was meant to
address, and present an alternative interpretation. We argue that maximum
likelihood is an inappropriate training objective when the end-goal is to
generate natural-looking samples. We go on to derive an ideal objective
function to use in this situation instead. We introduce a generalisation of
adversarial training, and show how such method can interpolate between maximum
likelihood training and our ideal training objective. To our knowledge this is
the first theoretical analysis that explains why adversarial training tends to
produce samples with higher perceived quality.
|
Resolving the Geometric Locus Dilemma for Support Vector Learning
Machines | Capacity control, the bias/variance dilemma, and learning unknown functions
from data, are all concerned with identifying effective and consistent fits of
unknown geometric loci to random data points. A geometric locus is a curve or
surface formed by points, all of which possess some uniform property. A
geometric locus of an algebraic equation is the set of points whose coordinates
are solutions of the equation. Any given curve or surface must pass through
each point on a specified locus. This paper argues that it is impossible to fit
random data points to algebraic equations of partially configured geometric
loci that reference arbitrary Cartesian coordinate systems. It also argues that
the fundamental curve of a linear decision boundary is actually a principal
eigenaxis. It is shown that learning principal eigenaxes of linear decision
boundaries involves finding a point of statistical equilibrium for which
eigenenergies of principal eigenaxis components are symmetrically balanced with
each other. It is demonstrated that learning linear decision boundaries
involves strong duality relationships between a statistical eigenlocus of
principal eigenaxis components and its algebraic forms, in primal and dual,
correlated Hilbert spaces. Locus equations are introduced and developed that
describe principal eigen-coordinate systems for lines, planes, and hyperplanes.
These equations are used to introduce and develop primal and dual statistical
eigenlocus equations of principal eigenaxes of linear decision boundaries.
Important generalizations for linear decision boundaries are shown to be
encoded within a dual statistical eigenlocus of principal eigenaxis components.
Principal eigenaxes of linear decision boundaries are shown to encode Bayes'
likelihood ratio for common covariance data and a robust likelihood ratio for
all other data.
|
Random sampling of bandlimited signals on graphs | We study the problem of sampling k-bandlimited signals on graphs. We propose
two sampling strategies that consist in selecting a small subset of nodes at
random. The first strategy is non-adaptive, i.e., independent of the graph
structure, and its performance depends on a parameter called the graph
coherence. On the contrary, the second strategy is adaptive but yields optimal
results. Indeed, no more than O(k log(k)) measurements are sufficient to ensure
an accurate and stable recovery of all k-bandlimited signals. This second
strategy is based on a careful choice of the sampling distribution, which can
be estimated quickly. Then, we propose a computationally efficient decoder to
reconstruct k-bandlimited signals from their samples. We prove that it yields
accurate reconstructions and that it is also stable to noise. Finally, we
conduct several experiments to test these techniques.
|
Deep Kalman Filters | Kalman Filters are one of the most influential models of time-varying
phenomena. They admit an intuitive probabilistic interpretation, have a simple
functional form, and enjoy widespread adoption in a variety of disciplines.
Motivated by recent variational methods for learning deep generative models, we
introduce a unified algorithm to efficiently learn a broad spectrum of Kalman
filters. Of particular interest is the use of temporal generative models for
counterfactual inference. We investigate the efficacy of such models for
counterfactual inference, and to that end we introduce the "Healing MNIST"
dataset where long-term structure, noise and actions are applied to sequences
of digits. We show the efficacy of our method for modeling this dataset. We
further show how our model can be used for counterfactual inference for
patients, based on electronic health record data of 8,000 patients over 4.5
years.
|
Adversarial Manipulation of Deep Representations | We show that the representation of an image in a deep neural network (DNN)
can be manipulated to mimic those of other natural images, with only minor,
imperceptible perturbations to the original image. Previous methods for
generating adversarial images focused on image perturbations designed to
produce erroneous class labels, while we concentrate on the internal layers of
DNN representations. In this way our new class of adversarial images differs
qualitatively from others. While the adversary is perceptually similar to one
image, its internal representation appears remarkably similar to a different
image, one from a different class, bearing little if any apparent similarity to
the input; they appear generic and consistent with the space of natural images.
This phenomenon raises questions about DNN representations, as well as the
properties of natural images themselves.
|
Fast Proximal Linearized Alternating Direction Method of Multiplier with
Parallel Splitting | The Augmented Lagragian Method (ALM) and Alternating Direction Method of
Multiplier (ADMM) have been powerful optimization methods for general convex
programming subject to linear constraint. We consider the convex problem whose
objective consists of a smooth part and a nonsmooth but simple part. We propose
the Fast Proximal Augmented Lagragian Method (Fast PALM) which achieves the
convergence rate $O(1/K^2)$, compared with $O(1/K)$ by the traditional PALM. In
order to further reduce the per-iteration complexity and handle the
multi-blocks problem, we propose the Fast Proximal ADMM with Parallel Splitting
(Fast PL-ADMM-PS) method. It also partially improves the rate related to the
smooth part of the objective function. Experimental results on both synthesized
and real world data demonstrate that our fast methods significantly improve the
previous PALM and ADMM.
|
Convolutional Models for Joint Object Categorization and Pose Estimation | In the task of Object Recognition, there exists a dichotomy between the
categorization of objects and estimating object pose, where the former
necessitates a view-invariant representation, while the latter requires a
representation capable of capturing pose information over different categories
of objects. With the rise of deep architectures, the prime focus has been on
object category recognition. Deep learning methods have achieved wide success
in this task. In contrast, object pose regression using these approaches has
received relatively much less attention. In this paper we show how deep
architectures, specifically Convolutional Neural Networks (CNN), can be adapted
to the task of simultaneous categorization and pose estimation of objects. We
investigate and analyze the layers of various CNN models and extensively
compare between them with the goal of discovering how the layers of distributed
representations of CNNs represent object pose information and how this
contradicts with object category representations. We extensively experiment on
two recent large and challenging multi-view datasets. Our models achieve better
than state-of-the-art performance on both datasets.
|
MuProp: Unbiased Backpropagation for Stochastic Neural Networks | Deep neural networks are powerful parametric models that can be trained
efficiently using the backpropagation algorithm. Stochastic neural networks
combine the power of large parametric functions with that of graphical models,
which makes it possible to learn very complex distributions. However, as
backpropagation is not directly applicable to stochastic networks that include
discrete sampling operations within their computational graph, training such
networks remains difficult. We present MuProp, an unbiased gradient estimator
for stochastic networks, designed to make this task easier. MuProp improves on
the likelihood-ratio estimator by reducing its variance using a control variate
based on the first-order Taylor expansion of a mean-field network. Crucially,
unlike prior attempts at using backpropagation for training stochastic
networks, the resulting estimator is unbiased and well behaved. Our experiments
on structured output prediction and discrete latent variable modeling
demonstrate that MuProp yields consistently good performance across a range of
difficult tasks.
|
Binary Classifier Calibration using an Ensemble of Near Isotonic
Regression Models | Learning accurate probabilistic models from data is crucial in many practical
tasks in data mining. In this paper we present a new non-parametric calibration
method called \textit{ensemble of near isotonic regression} (ENIR). The method
can be considered as an extension of BBQ, a recently proposed calibration
method, as well as the commonly used calibration method based on isotonic
regression. ENIR is designed to address the key limitation of isotonic
regression which is the monotonicity assumption of the predictions. Similar to
BBQ, the method post-processes the output of a binary classifier to obtain
calibrated probabilities. Thus it can be combined with many existing
classification models. We demonstrate the performance of ENIR on synthetic and
real datasets for the commonly used binary classification models. Experimental
results show that the method outperforms several common binary classifier
calibration methods. In particular on the real data, ENIR commonly performs
statistically significantly better than the other methods, and never worse. It
is able to improve the calibration power of classifiers, while retaining their
discrimination power. The method is also computationally tractable for large
scale datasets, as it is $O(N \log N)$ time, where $N$ is the number of
samples.
|
Sparse-promoting Full Waveform Inversion based on Online Orthonormal
Dictionary Learning | Full waveform inversion (FWI) delivers high-resolution images of the
subsurface by minimizing iteratively the misfit between the recorded and
calculated seismic data. It has been attacked successfully with the
Gauss-Newton method and sparsity promoting regularization based on fixed
multiscale transforms that permit significant subsampling of the seismic data
when the model perturbation at each FWI data-fitting iteration can be
represented with sparse coefficients. Rather than using analytical transforms
with predefined dictionaries to achieve sparse representation, we introduce an
adaptive transform called the Sparse Orthonormal Transform (SOT) whose
dictionary is learned from many small training patches taken from the model
perturbations in previous iterations. The patch-based dictionary is constrained
to be orthonormal and trained with an online approach to provide the best
sparse representation of the complex features and variations of the entire
model perturbation. The complexity of the training method is proportional to
the cube of the number of samples in one small patch. By incorporating both
compressive subsampling and the adaptive SOT-based representation into the
Gauss-Newton least-squares problem for each FWI iteration, the model
perturbation can be recovered after an l1-norm sparsity constraint is applied
on the SOT coefficients. Numerical experiments on synthetic models demonstrate
that the SOT-based sparsity promoting regularization can provide robust FWI
results with reduced computation.
|
Binary embeddings with structured hashed projections | We consider the hashing mechanism for constructing binary embeddings, that
involves pseudo-random projections followed by nonlinear (sign function)
mappings. The pseudo-random projection is described by a matrix, where not all
entries are independent random variables but instead a fixed "budget of
randomness" is distributed across the matrix. Such matrices can be efficiently
stored in sub-quadratic or even linear space, provide reduction in randomness
usage (i.e. number of required random values), and very often lead to
computational speed ups. We prove several theoretical results showing that
projections via various structured matrices followed by nonlinear mappings
accurately preserve the angular distance between input high-dimensional
vectors. To the best of our knowledge, these results are the first that give
theoretical ground for the use of general structured matrices in the nonlinear
setting. In particular, they generalize previous extensions of the
Johnson-Lindenstrauss lemma and prove the plausibility of the approach that was
so far only heuristically confirmed for some special structured matrices.
Consequently, we show that many structured matrices can be used as an efficient
information compression mechanism. Our findings build a better understanding of
certain deep architectures, which contain randomly weighted and untrained
layers, and yet achieve high performance on different learning tasks. We
empirically verify our theoretical findings and show the dependence of learning
via structured hashed projections on the performance of neural network as well
as nearest neighbor classifier.
|
How much does your data exploration overfit? Controlling bias via
information usage | Modern data is messy and high-dimensional, and it is often not clear a priori
what are the right questions to ask. Instead, the analyst typically needs to
use the data to search for interesting analyses to perform and hypotheses to
test. This is an adaptive process, where the choice of analysis to be performed
next depends on the results of the previous analyses on the same data.
Ultimately, which results are reported can be heavily influenced by the data.
It is widely recognized that this process, even if well-intentioned, can lead
to biases and false discoveries, contributing to the crisis of reproducibility
in science. But while %the adaptive nature of exploration any data-exploration
renders standard statistical theory invalid, experience suggests that different
types of exploratory analysis can lead to disparate levels of bias, and the
degree of bias also depends on the particulars of the data set. In this paper,
we propose a general information usage framework to quantify and provably bound
the bias and other error metrics of an arbitrary exploratory analysis. We prove
that our mutual information based bound is tight in natural settings, and then
use it to give rigorous insights into when commonly used procedures do or do
not lead to substantially biased estimation. Through the lens of information
usage, we analyze the bias of specific exploration procedures such as
filtering, rank selection and clustering. Our general framework also naturally
motivates randomization techniques that provably reduces exploration bias while
preserving the utility of the data analysis. We discuss the connections between
our approach and related ideas from differential privacy and blinded data
analysis, and supplement our results with illustrative simulations.
|
Reduced-Precision Strategies for Bounded Memory in Deep Neural Nets | This work investigates how using reduced precision data in Convolutional
Neural Networks (CNNs) affects network accuracy during classification. More
specifically, this study considers networks where each layer may use different
precision data. Our key result is the observation that the tolerance of CNNs to
reduced precision data not only varies across networks, a well established
observation, but also within networks. Tuning precision per layer is appealing
as it could enable energy and performance improvements. In this paper we study
how error tolerance across layers varies and propose a method for finding a low
precision configuration for a network while maintaining high accuracy. A
diverse set of CNNs is analyzed showing that compared to a conventional
implementation using a 32-bit floating-point representation for all layers, and
with less than 1% loss in relative accuracy, the data footprint required by
these networks can be reduced by an average of 74% and up to 92%.
|
An extension of McDiarmid's inequality | We derive an extension of McDiarmid's inequality for functions $f$ with
bounded differences on a high probability set ${\cal Y}$ (instead of almost
surely). The behavior of $f$ outside ${\cal Y}$ may be arbitrary. The proof is
short and elementary, and relies on an extension argument similar to
Kirszbraun's theorem.
|
Robust PCA via Nonconvex Rank Approximation | Numerous applications in data mining and machine learning require recovering
a matrix of minimal rank. Robust principal component analysis (RPCA) is a
general framework for handling this kind of problems. Nuclear norm based convex
surrogate of the rank function in RPCA is widely investigated. Under certain
assumptions, it can recover the underlying true low rank matrix with high
probability. However, those assumptions may not hold in real-world
applications. Since the nuclear norm approximates the rank by adding all
singular values together, which is essentially a $\ell_1$-norm of the singular
values, the resulting approximation error is not trivial and thus the resulting
matrix estimator can be significantly biased. To seek a closer approximation
and to alleviate the above-mentioned limitations of the nuclear norm, we
propose a nonconvex rank approximation. This approximation to the matrix rank
is tighter than the nuclear norm. To solve the associated nonconvex
minimization problem, we develop an efficient augmented Lagrange multiplier
based optimization algorithm. Experimental results demonstrate that our method
outperforms current state-of-the-art algorithms in both accuracy and
efficiency.
|
The Use of Machine Learning Algorithms in Recommender Systems: A
Systematic Review | Recommender systems use algorithms to provide users with product or service
recommendations. Recently, these systems have been using machine learning
algorithms from the field of artificial intelligence. However, choosing a
suitable machine learning algorithm for a recommender system is difficult
because of the number of algorithms described in the literature. Researchers
and practitioners developing recommender systems are left with little
information about the current approaches in algorithm usage. Moreover, the
development of a recommender system using a machine learning algorithm often
has problems and open questions that must be evaluated, so software engineers
know where to focus research efforts. This paper presents a systematic review
of the literature that analyzes the use of machine learning algorithms in
recommender systems and identifies research opportunities for software
engineering research. The study concludes that Bayesian and decision tree
algorithms are widely used in recommender systems because of their relative
simplicity, and that requirement and design phases of recommender system
development appear to offer opportunities for further research.
|
AUC-maximized Deep Convolutional Neural Fields for Sequence Labeling | Deep Convolutional Neural Networks (DCNN) has shown excellent performance in
a variety of machine learning tasks. This manuscript presents Deep
Convolutional Neural Fields (DeepCNF), a combination of DCNN with Conditional
Random Field (CRF), for sequence labeling with highly imbalanced label
distribution. The widely-used training methods, such as maximum-likelihood and
maximum labelwise accuracy, do not work well on highly imbalanced data. To
handle this, we present a new training algorithm called maximum-AUC for
DeepCNF. That is, we train DeepCNF by directly maximizing the empirical Area
Under the ROC Curve (AUC), which is an unbiased measurement for imbalanced
data. To fulfill this, we formulate AUC in a pairwise ranking framework,
approximate it by a polynomial function and then apply a gradient-based
procedure to optimize it. We then test our AUC-maximized DeepCNF on three very
different protein sequence labeling tasks: solvent accessibility prediction,
8-state secondary structure prediction, and disorder prediction. Our
experimental results confirm that maximum-AUC greatly outperforms the other two
training methods on 8-state secondary structure prediction and disorder
prediction since their label distributions are highly imbalanced and also have
similar performance as the other two training methods on the solvent
accessibility prediction problem which has three equally-distributed labels.
Furthermore, our experimental results also show that our AUC-trained DeepCNF
models greatly outperform existing popular predictors of these three tasks.
|
Semi-supervised Collaborative Ranking with Push at Top | Existing collaborative ranking based recommender systems tend to perform best
when there is enough observed ratings for each user and the observation is made
completely at random. Under this setting recommender systems can properly
suggest a list of recommendations according to the user interests. However,
when the observed ratings are extremely sparse (e.g. in the case of cold-start
users where no rating data is available), and are not sampled uniformly at
random, existing ranking methods fail to effectively leverage side information
to transduct the knowledge from existing ratings to unobserved ones. We propose
a semi-supervised collaborative ranking model, dubbed \texttt{S$^2$COR}, to
improve the quality of cold-start recommendation. \texttt{S$^2$COR} mitigates
the sparsity issue by leveraging side information about both observed and
missing ratings by collaboratively learning the ranking model. This enables it
to deal with the case of missing data not at random, but to also effectively
incorporate the available side information in transduction. We experimentally
evaluated our proposed algorithm on a number of challenging real-world datasets
and compared against state-of-the-art models for cold-start recommendation. We
report significantly higher quality recommendations with our algorithm compared
to the state-of-the-art.
|
On the interplay of network structure and gradient convergence in deep
learning | The regularization and output consistency behavior of dropout and layer-wise
pretraining for learning deep networks have been fairly well studied. However,
our understanding of how the asymptotic convergence of backpropagation in deep
architectures is related to the structural properties of the network and other
design choices (like denoising and dropout rate) is less clear at this time. An
interesting question one may ask is whether the network architecture and input
data statistics may guide the choices of learning parameters and vice versa. In
this work, we explore the association between such structural, distributional
and learnability aspects vis-\`a-vis their interaction with parameter
convergence rates. We present a framework to address these questions based on
convergence of backpropagation for general nonconvex objectives using
first-order information. This analysis suggests an interesting relationship
between feature denoising and dropout. Building upon these results, we obtain a
setup that provides systematic guidance regarding the choice of learning
parameters and network sizes that achieve a certain level of convergence (in
the optimization sense) often mediated by statistical attributes of the inputs.
Our results are supported by a set of experimental evaluations as well as
independent empirical observations reported by other groups.
|
Structural-RNN: Deep Learning on Spatio-Temporal Graphs | Deep Recurrent Neural Network architectures, though remarkably capable at
modeling sequences, lack an intuitive high-level spatio-temporal structure.
That is while many problems in computer vision inherently have an underlying
high-level structure and can benefit from it. Spatio-temporal graphs are a
popular tool for imposing such high-level intuitions in the formulation of real
world problems. In this paper, we propose an approach for combining the power
of high-level spatio-temporal graphs and sequence learning success of Recurrent
Neural Networks~(RNNs). We develop a scalable method for casting an arbitrary
spatio-temporal graph as a rich RNN mixture that is feedforward, fully
differentiable, and jointly trainable. The proposed method is generic and
principled as it can be used for transforming any spatio-temporal graph through
employing a certain set of well defined steps. The evaluations of the proposed
approach on a diverse set of problems, ranging from modeling human motion to
object interactions, shows improvement over the state-of-the-art with a large
margin. We expect this method to empower new approaches to problem formulation
through high-level spatio-temporal graphs and Recurrent Neural Networks.
|
Constant Time EXPected Similarity Estimation using Stochastic
Optimization | A new algorithm named EXPected Similarity Estimation (EXPoSE) was recently
proposed to solve the problem of large-scale anomaly detection. It is a
non-parametric and distribution free kernel method based on the Hilbert space
embedding of probability measures. Given a dataset of $n$ samples, EXPoSE needs
only $\mathcal{O}(n)$ (linear time) to build a model and $\mathcal{O}(1)$
(constant time) to make a prediction. In this work we improve the linear
computational complexity and show that an $\epsilon$-accurate model can be
estimated in constant time, which has significant implications for large-scale
learning problems. To achieve this goal, we cast the original EXPoSE
formulation into a stochastic optimization problem. It is crucial that this
approach allows us to determine the number of iteration based on a desired
accuracy $\epsilon$, independent of the dataset size $n$. We will show that the
proposed stochastic gradient descent algorithm works in general (possible
infinite-dimensional) Hilbert spaces, is easy to implement and requires no
additional step-size parameters.
|
Bayesian Optimization with Dimension Scheduling: Application to
Biological Systems | Bayesian Optimization (BO) is a data-efficient method for global black-box
optimization of an expensive-to-evaluate fitness function. BO typically assumes
that computation cost of BO is cheap, but experiments are time consuming or
costly. In practice, this allows us to optimize ten or fewer critical
parameters in up to 1,000 experiments. But experiments may be less expensive
than BO methods assume: In some simulation models, we may be able to conduct
multiple thousands of experiments in a few hours, and the computational burden
of BO is no longer negligible compared to experimentation time. To address this
challenge we introduce a new Dimension Scheduling Algorithm (DSA), which
reduces the computational burden of BO for many experiments. The key idea is
that DSA optimizes the fitness function only along a small set of dimensions at
each iteration. This DSA strategy (1) reduces the necessary computation time,
(2) finds good solutions faster than the traditional BO method, and (3) can be
parallelized straightforwardly. We evaluate the DSA in the context of
optimizing parameters of dynamic models of microalgae metabolism and show
faster convergence than traditional BO.
|
Learning the Dimensionality of Word Embeddings | We describe a method for learning word embeddings with data-dependent
dimensionality. Our Stochastic Dimensionality Skip-Gram (SD-SG) and Stochastic
Dimensionality Continuous Bag-of-Words (SD-CBOW) are nonparametric analogs of
Mikolov et al.'s (2013) well-known 'word2vec' models. Vector dimensionality is
made dynamic by employing techniques used by Cote & Larochelle (2016) to define
an RBM with an infinite number of hidden units. We show qualitatively and
quantitatively that SD-SG and SD-CBOW are competitive with their
fixed-dimension counterparts while providing a distribution over embedding
dimensionalities, which offers a window into how semantics distribute across
dimensions.
|
Understanding Adversarial Training: Increasing Local Stability of Neural
Nets through Robust Optimization | We propose a general framework for increasing local stability of Artificial
Neural Nets (ANNs) using Robust Optimization (RO). We achieve this through an
alternating minimization-maximization procedure, in which the loss of the
network is minimized over perturbed examples that are generated at each
parameter update. We show that adversarial training of ANNs is in fact
robustification of the network optimization, and that our proposed framework
generalizes previous approaches for increasing local stability of ANNs.
Experimental results reveal that our approach increases the robustness of the
network to existing adversarial examples, while making it harder to generate
new ones. Furthermore, our algorithm improves the accuracy of the network also
on the original test data.
|
Deep multi-scale video prediction beyond mean square error | Learning to predict future images from a video sequence involves the
construction of an internal representation that models the image evolution
accurately, and therefore, to some degree, its content and dynamics. This is
why pixel-space video prediction may be viewed as a promising avenue for
unsupervised feature learning. In addition, while optical flow has been a very
studied problem in computer vision for a long time, future frame prediction is
rarely approached. Still, many vision applications could benefit from the
knowledge of the next frames of videos, that does not require the complexity of
tracking every pixel trajectories. In this work, we train a convolutional
network to generate future frames given an input sequence. To deal with the
inherently blurry predictions obtained from the standard Mean Squared Error
(MSE) loss function, we propose three different and complementary feature
learning strategies: a multi-scale architecture, an adversarial training
method, and an image gradient difference loss function. We compare our
predictions to different published results based on recurrent neural networks
on the UCF101 dataset
|
Extending Gossip Algorithms to Distributed Estimation of U-Statistics | Efficient and robust algorithms for decentralized estimation in networks are
essential to many distributed systems. Whereas distributed estimation of sample
mean statistics has been the subject of a good deal of attention, computation
of $U$-statistics, relying on more expensive averaging over pairs of
observations, is a less investigated area. Yet, such data functionals are
essential to describe global properties of a statistical population, with
important examples including Area Under the Curve, empirical variance, Gini
mean difference and within-cluster point scatter. This paper proposes new
synchronous and asynchronous randomized gossip algorithms which simultaneously
propagate data across the network and maintain local estimates of the
$U$-statistic of interest. We establish convergence rate bounds of $O(1/t)$ and
$O(\log t / t)$ for the synchronous and asynchronous cases respectively, where
$t$ is the number of iterations, with explicit data and network dependent
terms. Beyond favorable comparisons in terms of rate analysis, numerical
experiments provide empirical evidence the proposed algorithms surpasses the
previously introduced approach.
|
Gated Graph Sequence Neural Networks | Graph-structured data appears frequently in domains including chemistry,
natural language semantics, social networks, and knowledge bases. In this work,
we study feature learning techniques for graph-structured inputs. Our starting
point is previous work on Graph Neural Networks (Scarselli et al., 2009), which
we modify to use gated recurrent units and modern optimization techniques and
then extend to output sequences. The result is a flexible and broadly useful
class of neural network models that has favorable inductive biases relative to
purely sequence-based models (e.g., LSTMs) when the problem is
graph-structured. We demonstrate the capabilities on some simple AI (bAbI) and
graph algorithm learning tasks. We then show it achieves state-of-the-art
performance on a problem from program verification, in which subgraphs need to
be matched to abstract data structures.
|
Learning Neural Network Architectures using Backpropagation | Deep neural networks with millions of parameters are at the heart of many
state of the art machine learning models today. However, recent works have
shown that models with much smaller number of parameters can also perform just
as well. In this work, we introduce the problem of architecture-learning, i.e;
learning the architecture of a neural network along with weights. We introduce
a new trainable parameter called tri-state ReLU, which helps in eliminating
unnecessary neurons. We also propose a smooth regularizer which encourages the
total number of neurons after elimination to be small. The resulting objective
is differentiable and simple to optimize. We experimentally validate our method
on both small and large networks, and show that it can learn models with a
considerably small number of parameters without affecting prediction accuracy.
|
Automatic Instrument Recognition in Polyphonic Music Using Convolutional
Neural Networks | Traditional methods to tackle many music information retrieval tasks
typically follow a two-step architecture: feature engineering followed by a
simple learning algorithm. In these "shallow" architectures, feature
engineering and learning are typically disjoint and unrelated. Additionally,
feature engineering is difficult, and typically depends on extensive domain
expertise.
In this paper, we present an application of convolutional neural networks for
the task of automatic musical instrument identification. In this model, feature
extraction and learning algorithms are trained together in an end-to-end
fashion. We show that a convolutional neural network trained on raw audio can
achieve performance surpassing traditional methods that rely on hand-crafted
features.
|
Return of Frustratingly Easy Domain Adaptation | Unlike human learning, machine learning often fails to handle changes between
training (source) and test (target) input distributions. Such domain shifts,
common in practical scenarios, severely damage the performance of conventional
machine learning methods. Supervised domain adaptation methods have been
proposed for the case when the target data have labels, including some that
perform very well despite being "frustratingly easy" to implement. However, in
practice, the target domain is often unlabeled, requiring unsupervised
adaptation. We propose a simple, effective, and efficient method for
unsupervised domain adaptation called CORrelation ALignment (CORAL). CORAL
minimizes domain shift by aligning the second-order statistics of source and
target distributions, without requiring any target labels. Even though it is
extraordinarily simple--it can be implemented in four lines of Matlab
code--CORAL performs remarkably well in extensive evaluations on standard
benchmark datasets.
|
Identifying the Absorption Bump with Deep Learning | The pervasive interstellar dust grains provide significant insights to
understand the formation and evolution of the stars, planetary systems, and the
galaxies, and may harbor the building blocks of life. One of the most effective
way to analyze the dust is via their interaction with the light from background
sources. The observed extinction curves and spectral features carry the size
and composition information of dust. The broad absorption bump at 2175 Angstrom
is the most prominent feature in the extinction curves. Traditionally,
statistical methods are applied to detect the existence of the absorption bump.
These methods require heavy preprocessing and the co-existence of other
reference features to alleviate the influence from the noises. In this paper,
we apply Deep Learning techniques to detect the broad absorption bump. We
demonstrate the key steps for training the selected models and their results.
The success of Deep Learning based method inspires us to generalize a common
methodology for broader science discovery problems. We present our on-going
work to build the DeepDis system for such kind of applications.
|
A Block Regression Model for Short-Term Mobile Traffic Forecasting | Accurate mobile traffic forecast is important for efficient network planning
and operations. However, existing traffic forecasting models have high
complexity, making the forecasting process slow and costly. In this paper, we
analyze some characteristics of mobile traffic such as periodicity, spatial
similarity and short term relativity. Based on these characteristics, we
propose a \emph{Block Regression} ({BR}) model for mobile traffic forecasting.
This model employs seasonal differentiation so as to take into account of the
temporally repetitive nature of mobile traffic. One of the key features of our
{BR} model lies in its low complexity since it constructs a single model for
all base stations. We evaluate the accuracy of {BR} model based on real traffic
data and compare it with the existing models. Results show that our {BR} model
offers equal accuracy to the existing models but has much less complexity.
|
Learning Structured Inference Neural Networks with Label Relations | Images of scenes have various objects as well as abundant attributes, and
diverse levels of visual categorization are possible. A natural image could be
assigned with fine-grained labels that describe major components,
coarse-grained labels that depict high level abstraction or a set of labels
that reveal attributes. Such categorization at different concept layers can be
modeled with label graphs encoding label information. In this paper, we exploit
this rich information with a state-of-art deep learning framework, and propose
a generic structured model that leverages diverse label relations to improve
image classification performance. Our approach employs a novel stacked label
prediction neural network, capturing both inter-level and intra-level label
semantics. We evaluate our method on benchmark image datasets, and empirical
results illustrate the efficacy of our model.
|
Predicting distributions with Linearizing Belief Networks | Conditional belief networks introduce stochastic binary variables in neural
networks. Contrary to a classical neural network, a belief network can predict
more than the expected value of the output $Y$ given the input $X$. It can
predict a distribution of outputs $Y$ which is useful when an input can admit
multiple outputs whose average is not necessarily a valid answer. Such networks
are particularly relevant to inverse problems such as image prediction for
denoising, or text to speech. However, traditional sigmoid belief networks are
hard to train and are not suited to continuous problems. This work introduces a
new family of networks called linearizing belief nets or LBNs. A LBN decomposes
into a deep linear network where each linear unit can be turned on or off by
non-deterministic binary latent units. It is a universal approximator of
real-valued conditional distributions and can be trained using gradient
descent. Moreover, the linear pathways efficiently propagate continuous
information and they act as multiplicative skip-connections that help
optimization by removing gradient diffusion. This yields a model which trains
efficiently and improves the state-of-the-art on image denoising and facial
expression generation with the Toronto faces dataset.
|
Competitive Multi-scale Convolution | In this paper, we introduce a new deep convolutional neural network (ConvNet)
module that promotes competition among a set of multi-scale convolutional
filters. This new module is inspired by the inception module, where we replace
the original collaborative pooling stage (consisting of a concatenation of the
multi-scale filter outputs) by a competitive pooling represented by a maxout
activation unit. This extension has the following two objectives: 1) the
selection of the maximum response among the multi-scale filters prevents filter
co-adaptation and allows the formation of multiple sub-networks within the same
model, which has been shown to facilitate the training of complex learning
problems; and 2) the maxout unit reduces the dimensionality of the outputs from
the multi-scale filters. We show that the use of our proposed module in typical
deep ConvNets produces classification results that are either better than or
comparable to the state of the art on the following benchmark datasets: MNIST,
CIFAR-10, CIFAR-100 and SVHN.
|
Net2Net: Accelerating Learning via Knowledge Transfer | We introduce techniques for rapidly transferring the information stored in
one neural net into another neural net. The main purpose is to accelerate the
training of a significantly larger neural net. During real-world workflows, one
often trains very many different neural networks during the experimentation and
design process. This is a wasteful process in which each new model is trained
from scratch. Our Net2Net technique accelerates the experimentation process by
instantaneously transferring the knowledge from a previous network to each new
deeper or wider network. Our techniques are based on the concept of
function-preserving transformations between neural network specifications. This
differs from previous approaches to pre-training that altered the function
represented by a neural net when adding layers to it. Using our knowledge
transfer mechanism to add depth to Inception modules, we demonstrate a new
state of the art accuracy rating on the ImageNet dataset.
|
A New Smooth Approximation to the Zero One Loss with a Probabilistic
Interpretation | We examine a new form of smooth approximation to the zero one loss in which
learning is performed using a reformulation of the widely used logistic
function. Our approach is based on using the posterior mean of a novel
generalized Beta-Bernoulli formulation. This leads to a generalized logistic
function that approximates the zero one loss, but retains a probabilistic
formulation conferring a number of useful properties. The approach is easily
generalized to kernel logistic regression and easily integrated into methods
for structured prediction. We present experiments in which we learn such models
using an optimization method consisting of a combination of gradient descent
and coordinate descent using localized grid search so as to escape from local
minima. Our experiments indicate that optimization quality is improved when
learning meta-parameters are themselves optimized using a validation set. Our
experiments show improved performance relative to widely used logistic and
hinge loss methods on a wide variety of problems ranging from standard UC
Irvine and libSVM evaluation datasets to product review predictions and a
visual information extraction task. We observe that the approach: 1) is more
robust to outliers compared to the logistic and hinge losses; 2) outperforms
comparable logistic and max margin models on larger scale benchmark problems;
3) when combined with Gaussian- Laplacian mixture prior on parameters the
kernelized version of our formulation yields sparser solutions than Support
Vector Machine classifiers; and 4) when integrated into a probabilistic
structured prediction technique our approach provides more accurate
probabilities yielding improved inference and increasing information extraction
performance.
|
Adversarial Autoencoders | In this paper, we propose the "adversarial autoencoder" (AAE), which is a
probabilistic autoencoder that uses the recently proposed generative
adversarial networks (GAN) to perform variational inference by matching the
aggregated posterior of the hidden code vector of the autoencoder with an
arbitrary prior distribution. Matching the aggregated posterior to the prior
ensures that generating from any part of prior space results in meaningful
samples. As a result, the decoder of the adversarial autoencoder learns a deep
generative model that maps the imposed prior to the data distribution. We show
how the adversarial autoencoder can be used in applications such as
semi-supervised classification, disentangling style and content of images,
unsupervised clustering, dimensionality reduction and data visualization. We
performed experiments on MNIST, Street View House Numbers and Toronto Face
datasets and show that adversarial autoencoders achieve competitive results in
generative modeling and semi-supervised classification tasks.
|
Tree-Guided MCMC Inference for Normalized Random Measure Mixture Models | Normalized random measures (NRMs) provide a broad class of discrete random
measures that are often used as priors for Bayesian nonparametric models.
Dirichlet process is a well-known example of NRMs. Most of posterior inference
methods for NRM mixture models rely on MCMC methods since they are easy to
implement and their convergence is well studied. However, MCMC often suffers
from slow convergence when the acceptance rate is low. Tree-based inference is
an alternative deterministic posterior inference method, where Bayesian
hierarchical clustering (BHC) or incremental Bayesian hierarchical clustering
(IBHC) have been developed for DP or NRM mixture (NRMM) models, respectively.
Although IBHC is a promising method for posterior inference for NRMM models due
to its efficiency and applicability to online inference, its convergence is not
guaranteed since it uses heuristics that simply selects the best solution after
multiple trials are made. In this paper, we present a hybrid inference
algorithm for NRMM models, which combines the merits of both MCMC and IBHC.
Trees built by IBHC outlines partitions of data, which guides
Metropolis-Hastings procedure to employ appropriate proposals. Inheriting the
nature of MCMC, our tree-guided MCMC (tgMCMC) is guaranteed to converge, and
enjoys the fast convergence thanks to the effective proposals guided by trees.
Experiments on both synthetic and real-world datasets demonstrate the benefit
of our method.
|
Why are deep nets reversible: A simple theory, with implications for
training | Generative models for deep learning are promising both to improve
understanding of the model, and yield training methods requiring fewer labeled
samples.
Recent works use generative model approaches to produce the deep net's input
given the value of a hidden layer several levels above. However, there is no
accompanying "proof of correctness" for the generative model, showing that the
feedforward deep net is the correct inference method for recovering the hidden
layer given the input. Furthermore, these models are complicated.
The current paper takes a more theoretical tack. It presents a very simple
generative model for RELU deep nets, with the following characteristics: (i)
The generative model is just the reverse of the feedforward net: if the forward
transformation at a layer is $A$ then the reverse transformation is $A^T$.
(This can be seen as an explanation of the old weight tying idea for denoising
autoencoders.) (ii) Its correctness can be proven under a clean theoretical
assumption: the edge weights in real-life deep nets behave like random numbers.
Under this assumption ---which is experimentally tested on real-life nets like
AlexNet--- it is formally proved that feed forward net is a correct inference
method for recovering the hidden layer.
The generative model suggests a simple modification for training: use the
generative model to produce synthetic data with labels and include it in the
training set. Experiments are shown to support this theory of random-like deep
nets; and that it helps the training.
|
Expressiveness of Rectifier Networks | Rectified Linear Units (ReLUs) have been shown to ameliorate the vanishing
gradient problem, allow for efficient backpropagation, and empirically promote
sparsity in the learned parameters. They have led to state-of-the-art results
in a variety of applications. However, unlike threshold and sigmoid networks,
ReLU networks are less explored from the perspective of their expressiveness.
This paper studies the expressiveness of ReLU networks. We characterize the
decision boundary of two-layer ReLU networks by constructing functionally
equivalent threshold networks. We show that while the decision boundary of a
two-layer ReLU network can be captured by a threshold network, the latter may
require an exponentially larger number of hidden units. We also formulate
sufficient conditions for a corresponding logarithmic reduction in the number
of hidden units to represent a sign network as a ReLU network. Finally, we
experimentally compare threshold networks and their much smaller ReLU
counterparts with respect to their ability to learn from synthetically
generated data.
|
A Distribution Adaptive Framework for Prediction Interval Estimation
Using Nominal Variables | Proposed methods for prediction interval estimation so far focus on cases
where input variables are numerical. In datasets with solely nominal input
variables, we observe records with the exact same input $x^u$, but different
real valued outputs due to the inherent noise in the system. Existing
prediction interval estimation methods do not use representations that can
accurately model such inherent noise in the case of nominal inputs. We propose
a new prediction interval estimation method tailored for this type of data,
which is prevalent in biology and medicine. We call this method Distribution
Adaptive Prediction Interval Estimation given Nominal inputs (DAPIEN) and has
four main phases. First, we select a distribution function that can best
represent the inherent noise of the system for all unique inputs. Then we infer
the parameters $\theta_i$ (e.g. $\theta_i=[mean_i, variance_i]$) of the
selected distribution function for all unique input vectors $x^u_i$ and
generate a new corresponding training set using pairs of $x^u_i, \theta_i$.
III). Then, we train a model to predict $\theta$ given a new $x_u$. Finally, we
calculate the prediction interval for a new sample using the inverse of the
cumulative distribution function once the parameters $\theta$ is predicted by
the trained model. We compared DAPIEN to the commonly used Bootstrap method on
three synthetic datasets. Our results show that DAPIEN provides tighter
prediction intervals while preserving the requested coverage when compared to
Bootstrap. This work can facilitate broader usage of regression methods in
medicine and biology where it is necessary to provide tight prediction
intervals while preserving coverage when input variables are nominal.
|
Efficient Output Kernel Learning for Multiple Tasks | The paradigm of multi-task learning is that one can achieve better
generalization by learning tasks jointly and thus exploiting the similarity
between the tasks rather than learning them independently of each other. While
previously the relationship between tasks had to be user-defined in the form of
an output kernel, recent approaches jointly learn the tasks and the output
kernel. As the output kernel is a positive semidefinite matrix, the resulting
optimization problems are not scalable in the number of tasks as an
eigendecomposition is required in each step. \mbox{Using} the theory of
positive semidefinite kernels we show in this paper that for a certain class of
regularizers on the output kernel, the constraint of being positive
semidefinite can be dropped as it is automatically satisfied for the relaxed
problem. This leads to an unconstrained dual problem which can be solved
efficiently. Experiments on several multi-task and multi-class data sets
illustrate the efficacy of our approach in terms of computational efficiency as
well as generalization performance.
|
Complex-Valued Gaussian Processes for Regression | In this paper we propose a novel Bayesian solution for nonlinear regression
in complex fields. Previous solutions for kernels methods usually assume a
complexification approach, where the real-valued kernel is replaced by a
complex-valued one. This approach is limited. Based on results in
complex-valued linear theory and Gaussian random processes we show that a
pseudo-kernel must be included. This is the starting point to develop the new
complex-valued formulation for Gaussian process for regression (CGPR). We face
the design of the covariance and pseudo-covariance based on a convolution
approach and for several scenarios. Just in the particular case where the
outputs are proper, the pseudo-kernel cancels. Also, the hyperparameters of the
covariance {can be learnt} maximizing the marginal likelihood using Wirtinger's
calculus and patterned complex-valued matrix derivatives. In the experiments
included, we show how CGPR successfully solve systems where real and imaginary
parts are correlated. Besides, we successfully solve the nonlinear channel
equalization problem by developing a recursive solution with basis removal. We
report remarkable improvements compared to previous solutions: a 2-4 dB
reduction of the MSE with {just a quarter} of the training samples used by
previous approaches.
|
Online learning in repeated auctions | Motivated by online advertising auctions, we consider repeated Vickrey
auctions where goods of unknown value are sold sequentially and bidders only
learn (potentially noisy) information about a good's value once it is
purchased. We adopt an online learning approach with bandit feedback to model
this problem and derive bidding strategies for two models: stochastic and
adversarial. In the stochastic model, the observed values of the goods are
random variables centered around the true value of the good. In this case,
logarithmic regret is achievable when competing against well behaved
adversaries. In the adversarial model, the goods need not be identical and we
simply compare our performance against that of the best fixed bid in hindsight.
We show that sublinear regret is also achievable in this case and prove
matching minimax lower bounds. To our knowledge, this is the first complete set
of strategies for bidders participating in auctions of this type.
|
Sparse learning of maximum likelihood model for optimization of complex
loss function | Traditional machine learning methods usually minimize a simple loss function
to learn a predictive model, and then use a complex performance measure to
measure the prediction performance. However, minimizing a simple loss function
cannot guarantee that an optimal performance. In this paper, we study the
problem of optimizing the complex performance measure directly to obtain a
predictive model. We proposed to construct a maximum likelihood model for this
problem, and to learn the model parameter, we minimize a com- plex loss
function corresponding to the desired complex performance measure. To optimize
the loss function, we approximate the upper bound of the complex loss. We also
propose impose the sparsity to the model parameter to obtain a sparse model. An
objective is constructed by combining the upper bound of the loss function and
the sparsity of the model parameter, and we develop an iterative algorithm to
minimize it by using the fast iterative shrinkage- thresholding algorithm
framework. The experiments on optimization on three different complex
performance measures, including F-score, receiver operating characteristic
curve, and recall precision curve break even point, over three real-world
applications, aircraft event recognition of civil aviation safety, in- trusion
detection in wireless mesh networks, and image classification, show the
advantages of the proposed method over state-of-the-art methods.
|
Image Question Answering using Convolutional Neural Network with Dynamic
Parameter Prediction | We tackle image question answering (ImageQA) problem by learning a
convolutional neural network (CNN) with a dynamic parameter layer whose weights
are determined adaptively based on questions. For the adaptive parameter
prediction, we employ a separate parameter prediction network, which consists
of gated recurrent unit (GRU) taking a question as its input and a
fully-connected layer generating a set of candidate weights as its output.
However, it is challenging to construct a parameter prediction network for a
large number of parameters in the fully-connected dynamic parameter layer of
the CNN. We reduce the complexity of this problem by incorporating a hashing
technique, where the candidate weights given by the parameter prediction
network are selected using a predefined hash function to determine individual
weights in the dynamic parameter layer. The proposed network---joint network
with the CNN for ImageQA and the parameter prediction network---is trained
end-to-end through back-propagation, where its weights are initialized using a
pre-trained CNN and GRU. The proposed algorithm illustrates the
state-of-the-art performance on all available public ImageQA benchmarks.
|
Metric learning approach for graph-based label propagation | The efficiency of graph-based semi-supervised algorithms depends on the graph
of instances on which they are applied. The instances are often in a vectorial
form before a graph linking them is built. The construction of the graph relies
on a metric over the vectorial space that help define the weight of the
connection between entities. The classic choice for this metric is usually a
distance measure or a similarity measure based on the euclidean norm. We claim
that in some cases the euclidean norm on the initial vectorial space might not
be the more appropriate to solve the task efficiently. We propose an algorithm
that aims at learning the most appropriate vectorial representation for
building a graph on which the task at hand is solved efficiently.
|
Censoring Representations with an Adversary | In practice, there are often explicit constraints on what representations or
decisions are acceptable in an application of machine learning. For example it
may be a legal requirement that a decision must not favour a particular group.
Alternatively it can be that that representation of data must not have
identifying information. We address these two related issues by learning
flexible representations that minimize the capability of an adversarial critic.
This adversary is trying to predict the relevant sensitive variable from the
representation, and so minimizing the performance of the adversary ensures
there is little or no information in the representation about the sensitive
variable. We demonstrate this adversarial approach on two problems: making
decisions free from discrimination and removing private information from
images. We formulate the adversarial model as a minimax problem, and optimize
that minimax objective using a stochastic gradient alternate min-max optimizer.
We demonstrate the ability to provide discriminant free representations for
standard test problems, and compare with previous state of the art methods for
fairness, showing statistically significant improvement across most cases. The
flexibility of this method is shown via a novel problem: removing annotations
from images, from unaligned training examples of annotated and unannotated
images, and with no a priori knowledge of the form of annotation provided to
the model.
|
Combining Neural Networks and Log-linear Models to Improve Relation
Extraction | The last decade has witnessed the success of the traditional feature-based
method on exploiting the discrete structures such as words or lexical patterns
to extract relations from text. Recently, convolutional and recurrent neural
networks has provided very effective mechanisms to capture the hidden
structures within sentences via continuous representations, thereby
significantly advancing the performance of relation extraction. The advantage
of convolutional neural networks is their capacity to generalize the
consecutive k-grams in the sentences while recurrent neural networks are
effective to encode long ranges of sentence context. This paper proposes to
combine the traditional feature-based method, the convolutional and recurrent
neural networks to simultaneously benefit from their advantages. Our systematic
evaluation of different network architectures and combination methods
demonstrates the effectiveness of this approach and results in the
state-of-the-art performance on the ACE 2005 and SemEval dataset.
|
On the Global Linear Convergence of Frank-Wolfe Optimization Variants | The Frank-Wolfe (FW) optimization algorithm has lately re-gained popularity
thanks in particular to its ability to nicely handle the structured constraints
appearing in machine learning applications. However, its convergence rate is
known to be slow (sublinear) when the solution lies at the boundary. A simple
less-known fix is to add the possibility to take 'away steps' during
optimization, an operation that importantly does not require a feasibility
oracle. In this paper, we highlight and clarify several variants of the
Frank-Wolfe optimization algorithm that have been successfully applied in
practice: away-steps FW, pairwise FW, fully-corrective FW and Wolfe's minimum
norm point algorithm, and prove for the first time that they all enjoy global
linear convergence, under a weaker condition than strong convexity of the
objective. The constant in the convergence rate has an elegant interpretation
as the product of the (classical) condition number of the function with a novel
geometric quantity that plays the role of a 'condition number' of the
constraint set. We provide pointers to where these algorithms have made a
difference in practice, in particular with the flow polytope, the marginal
polytope and the base polytope for submodular optimization.
|
Seeding K-Means using Method of Moments | K-means is one of the most widely used algorithms for clustering in Data
Mining applications, which attempts to minimize the sum of the square of the
Euclidean distance of the points in the clusters from the respective means of
the clusters. However, K-means suffers from local minima problem and is not
guaranteed to converge to the optimal cost. K-means++ tries to address the
problem by seeding the means using a distance-based sampling scheme. However,
seeding the means in K-means++ needs $O\left(K\right)$ sequential passes
through the entire dataset, and this can be very costly for large datasets.
Here we propose a method of seeding the initial means based on factorizations
of higher order moments for bounded data. Our method takes $O\left(1\right)$
passes through the entire dataset to extract the initial set of means, and its
final cost can be proven to be within $O(\sqrt{K})$ of the optimal cost. We
demonstrate the performance of our algorithm in comparison with the existing
algorithms on various benchmark datasets.
|
Metric Learning with Adaptive Density Discrimination | Distance metric learning (DML) approaches learn a transformation to a
representation space where distance is in correspondence with a predefined
notion of similarity. While such models offer a number of compelling benefits,
it has been difficult for these to compete with modern classification
algorithms in performance and even in feature extraction.
In this work, we propose a novel approach explicitly designed to address a
number of subtle yet important issues which have stymied earlier DML
algorithms. It maintains an explicit model of the distributions of the
different classes in representation space. It then employs this knowledge to
adaptively assess similarity, and achieve local discrimination by penalizing
class distribution overlap.
We demonstrate the effectiveness of this idea on several tasks. Our approach
achieves state-of-the-art classification results on a number of fine-grained
visual recognition datasets, surpassing the standard softmax classifier and
outperforming triplet loss by a relative margin of 30-40%. In terms of
computational performance, it alleviates training inefficiencies in the
traditional triplet loss, reaching the same error in 5-30 times fewer
iterations. Beyond classification, we further validate the saliency of the
learnt representations via their attribute concentration and hierarchy recovery
properties, achieving 10-25% relative gains on the softmax classifier and
25-50% on triplet loss in these tasks.
|
Doctor AI: Predicting Clinical Events via Recurrent Neural Networks | Leveraging large historical data in electronic health record (EHR), we
developed Doctor AI, a generic predictive model that covers observed medical
conditions and medication uses. Doctor AI is a temporal model using recurrent
neural networks (RNN) and was developed and applied to longitudinal time
stamped EHR data from 260K patients over 8 years. Encounter records (e.g.
diagnosis codes, medication codes or procedure codes) were input to RNN to
predict (all) the diagnosis and medication categories for a subsequent visit.
Doctor AI assesses the history of patients to make multilabel predictions (one
label for each diagnosis or medication category). Based on separate blind test
set evaluation, Doctor AI can perform differential diagnosis with up to 79%
recall@30, significantly higher than several baselines. Moreover, we
demonstrate great generalizability of Doctor AI by adapting the resulting
models from one institution to another without losing substantial accuracy.
|
Unitary-Group Invariant Kernels and Features from Transformed Unlabeled
Data | The study of representations invariant to common transformations of the data
is important to learning. Most techniques have focused on local approximate
invariance implemented within expensive optimization frameworks lacking
explicit theoretical guarantees. In this paper, we study kernels that are
invariant to the unitary group while having theoretical guarantees in
addressing practical issues such as (1) unavailability of transformed versions
of labelled data and (2) not observing all transformations. We present a
theoretically motivated alternate approach to the invariant kernel SVM. Unlike
previous approaches to the invariant SVM, the proposed formulation solves both
issues mentioned. We also present a kernel extension of a recent technique to
extract linear unitary-group invariant features addressing both issues and
extend some guarantees regarding invariance and stability. We present
experiments on the UCI ML datasets to illustrate and validate our methods.
|
ACDC: A Structured Efficient Linear Layer | The linear layer is one of the most pervasive modules in deep learning
representations. However, it requires $O(N^2)$ parameters and $O(N^2)$
operations. These costs can be prohibitive in mobile applications or prevent
scaling in many domains. Here, we introduce a deep, differentiable,
fully-connected neural network module composed of diagonal matrices of
parameters, $\mathbf{A}$ and $\mathbf{D}$, and the discrete cosine transform
$\mathbf{C}$. The core module, structured as $\mathbf{ACDC^{-1}}$, has $O(N)$
parameters and incurs $O(N log N )$ operations. We present theoretical results
showing how deep cascades of ACDC layers approximate linear layers. ACDC is,
however, a stand-alone module and can be used in combination with any other
types of module. In our experiments, we show that it can indeed be successfully
interleaved with ReLU modules in convolutional neural networks for image
recognition. Our experiments also study critical factors in the training of
these structured modules, including initialization and depth. Finally, this
paper also provides a connection between structured linear transforms used in
deep learning and the field of Fourier optics, illustrating how ACDC could in
principle be implemented with lenses and diffractive elements.
|
Staleness-aware Async-SGD for Distributed Deep Learning | Deep neural networks have been shown to achieve state-of-the-art performance
in several machine learning tasks. Stochastic Gradient Descent (SGD) is the
preferred optimization algorithm for training these networks and asynchronous
SGD (ASGD) has been widely adopted for accelerating the training of large-scale
deep networks in a distributed computing environment. However, in practice it
is quite challenging to tune the training hyperparameters (such as learning
rate) when using ASGD so as achieve convergence and linear speedup, since the
stability of the optimization algorithm is strongly influenced by the
asynchronous nature of parameter updates. In this paper, we propose a variant
of the ASGD algorithm in which the learning rate is modulated according to the
gradient staleness and provide theoretical guarantees for convergence of this
algorithm. Experimental verification is performed on commonly-used image
classification benchmarks: CIFAR10 and Imagenet to demonstrate the superior
effectiveness of the proposed approach, compared to SSGD (Synchronous SGD) and
the conventional ASGD algorithm.
|
Prioritized Experience Replay | Experience replay lets online reinforcement learning agents remember and
reuse experiences from the past. In prior work, experience transitions were
uniformly sampled from a replay memory. However, this approach simply replays
transitions at the same frequency that they were originally experienced,
regardless of their significance. In this paper we develop a framework for
prioritizing experience, so as to replay important transitions more frequently,
and therefore learn more efficiently. We use prioritized experience replay in
Deep Q-Networks (DQN), a reinforcement learning algorithm that achieved
human-level performance across many Atari games. DQN with prioritized
experience replay achieves a new state-of-the-art, outperforming DQN with
uniform replay on 41 out of 49 games.
|
A pilot study on the daily control capability of s-EMG prosthetic hands
by amputees | Surface electromyography is a valid tool to gather muscular contraction
signals from intact and amputated subjects. Electromyographic signals can be
used to control prosthetic devices in a noninvasive way distinguishing the
movements performed by the particular EMG electrodes activity. According to the
literature, several algorithms have been used to control prosthetic hands
through s-EMG signals. The main issue is to correctly classify the signals
acquired as the movement actually performed. This work presents a study on the
Support Vector Machine's performance in a short-time period, gained using two
different feature representation (Mean Absolute Value and Waveform Length) of
the sEMG signals. In particular, we paid close attention to the repeatability
problem, that is the capability to achieve a stable and satisfactory level of
accuracy in repeated experiments. Results on a limited setting are encouraging,
as they show an average accuracy above 73% even in the worst case scenario.
|
Studying the control of non invasive prosthetic hands over large time
spans | The electromyography (EMG) signal is the electrical manifestation of a
neuromuscular activation that provides access to physiological processes which
cause the muscle to generate force and produce movement. Non invasive
prostheses use such signals detected by the electrodes placed on the user's
stump, as input to generate hand posture movements according to the intentions
of the prosthesis wearer. The aim of this pilot study is to explore the
repeatability issue, i.e. the ability to classify 17 different hand postures,
represented by EMG signal, across a time span of days by a control algorithm.
Data collection experiments lasted four days and signals were collected from
the forearm of a single subject. We find that Support Vector Machine (SVM)
classification results are high enough to guarantee a correct classification of
more than 10 postures in each moment of the considered time span.
|
Regret Analysis of the Finite-Horizon Gittins Index Strategy for
Multi-Armed Bandits | I analyse the frequentist regret of the famous Gittins index strategy for
multi-armed bandits with Gaussian noise and a finite horizon. Remarkably it
turns out that this approach leads to finite-time regret guarantees comparable
to those available for the popular UCB algorithm. Along the way I derive
finite-time bounds on the Gittins index that are asymptotically exact and may
be of independent interest. I also discuss some computational issues and
present experimental results suggesting that a particular version of the
Gittins index strategy is a modest improvement on existing algorithms with
finite-time regret guarantees such as UCB and Thompson sampling.
|
Segmental Recurrent Neural Networks | We introduce segmental recurrent neural networks (SRNNs) which define, given
an input sequence, a joint probability distribution over segmentations of the
input and labelings of the segments. Representations of the input segments
(i.e., contiguous subsequences of the input) are computed by encoding their
constituent tokens using bidirectional recurrent neural nets, and these
"segment embeddings" are used to define compatibility scores with output
labels. These local compatibility scores are integrated using a global
semi-Markov conditional random field. Both fully supervised training -- in
which segment boundaries and labels are observed -- as well as partially
supervised training -- in which segment boundaries are latent -- are
straightforward. Experiments on handwriting recognition and joint Chinese word
segmentation/POS tagging show that, compared to models that do not explicitly
represent segments such as BIO tagging schemes and connectionist temporal
classification (CTC), SRNNs obtain substantially higher accuracies.
|
Neural Variational Inference for Text Processing | Recent advances in neural variational inference have spawned a renaissance in
deep latent variable models. In this paper we introduce a generic variational
inference framework for generative and conditional models of text. While
traditional variational methods derive an analytic approximation for the
intractable distributions over latent variables, here we construct an inference
network conditioned on the discrete text input to provide the variational
distribution. We validate this framework on two very different text modelling
applications, generative document modelling and supervised question answering.
Our neural variational document model combines a continuous stochastic document
representation with a bag-of-words generative model and achieves the lowest
reported perplexities on two standard test corpora. The neural answer selection
model employs a stochastic representation layer within an attention mechanism
to extract the semantics between a question and answer pair. On two question
answering benchmarks this model exceeds all previous published benchmarks.
|
What Objective Does Self-paced Learning Indeed Optimize? | Self-paced learning (SPL) is a recently raised methodology designed through
simulating the learning principle of humans/animals. A variety of SPL
realization schemes have been designed for different computer vision and
pattern recognition tasks, and empirically substantiated to be effective in
these applications. However, the investigation on its theoretical insight is
still a blank. To this issue, this study attempts to provide some new
theoretical understanding under the SPL scheme. Specifically, we prove that the
solving strategy on SPL accords with a majorization minimization algorithm
implemented on a latent objective function. Furthermore, we find that the loss
function contained in this latent objective has a similar configuration with
non-convex regularized penalty (NSPR) known in statistics and machine learning.
Such connection inspires us discovering more intrinsic relationship between SPL
regimes and NSPR forms, like SCAD, LOG and EXP. The robustness insight under
SPL can then be finely explained. We also analyze the capability of SPL on its
easy loss prior embedding property, and provide an insightful interpretation to
the effectiveness mechanism under previous SPL variations. Besides, we design a
group-partial-order loss prior, which is especially useful to weakly labeled
large-scale data processing tasks. Through applying SPL with this loss prior to
the FCVID dataset, which is currently one of the biggest manually annotated
video dataset, our method achieves state-of-the-art performance beyond previous
methods, which further helps supports the proposed theoretical arguments.
|
SparkNet: Training Deep Networks in Spark | Training deep networks is a time-consuming process, with networks for object
recognition often requiring multiple days to train. For this reason, leveraging
the resources of a cluster to speed up training is an important area of work.
However, widely-popular batch-processing computational frameworks like
MapReduce and Spark were not designed to support the asynchronous and
communication-intensive workloads of existing distributed deep learning
systems. We introduce SparkNet, a framework for training deep networks in
Spark. Our implementation includes a convenient interface for reading data from
Spark RDDs, a Scala interface to the Caffe deep learning framework, and a
lightweight multi-dimensional tensor library. Using a simple parallelization
scheme for stochastic gradient descent, SparkNet scales well with the cluster
size and tolerates very high-latency communication. Furthermore, it is easy to
deploy and use with no parameter tuning, and it is compatible with existing
Caffe models. We quantify the dependence of the speedup obtained by SparkNet on
the number of machines, the communication frequency, and the cluster's
communication overhead, and we benchmark our system's performance on the
ImageNet dataset.
|
A Novel Approach for Phase Identification in Smart Grids Using Graph
Theory and Principal Component Analysis | Consumers with low demand, like households, are generally supplied
single-phase power by connecting their service mains to one of the phases of a
distribution transformer. The distribution companies face the problem of
keeping a record of consumer connectivity to a phase due to uninformed changes
that happen. The exact phase connectivity information is important for the
efficient operation and control of distribution system. We propose a new data
driven approach to the problem based on Principal Component Analysis (PCA) and
its Graph Theoretic interpretations, using energy measurements in equally timed
short intervals, generated from smart meters. We propose an algorithm for
inferring phase connectivity from noisy measurements. The algorithm is
demonstrated using simulated data for phase connectivities in distribution
networks.
|
Deep Learning for Tactile Understanding From Visual and Haptic Data | Robots which interact with the physical world will benefit from a
fine-grained tactile understanding of objects and surfaces. Additionally, for
certain tasks, robots may need to know the haptic properties of an object
before touching it. To enable better tactile understanding for robots, we
propose a method of classifying surfaces with haptic adjectives (e.g.,
compressible or smooth) from both visual and physical interaction data. Humans
typically combine visual predictions and feedback from physical interactions to
accurately predict haptic properties and interact with the world. Inspired by
this cognitive pattern, we propose and explore a purely visual haptic
prediction model. Purely visual models enable a robot to "feel" without
physical interaction. Furthermore, we demonstrate that using both visual and
physical interaction signals together yields more accurate haptic
classification. Our models take advantage of recent advances in deep neural
networks by employing a unified approach to learning features for physical
interaction and visual observations. Even though we employ little domain
specific knowledge, our model still achieves better results than methods based
on hand-designed features.
|
Transfer Learning for Speech and Language Processing | Transfer learning is a vital technique that generalizes models trained for
one setting or task to other settings or tasks. For example in speech
recognition, an acoustic model trained for one language can be used to
recognize speech in another language, with little or no re-training data.
Transfer learning is closely related to multi-task learning (cross-lingual vs.
multilingual), and is traditionally studied in the name of `model adaptation'.
Recent advance in deep learning shows that transfer learning becomes much
easier and more effective with high-level abstract features learned by deep
models, and the `transfer' can be conducted not only between data distributions
and data types, but also between model structures (e.g., shallow nets and deep
nets) or even model types (e.g., Bayesian models and neural models). This
review paper summarizes some recent prominent research towards this direction,
particularly for speech and language processing. We also report some results
from our group and highlight the potential of this very interesting research
field.
|
Convolutional neural networks with low-rank regularization | Large CNNs have delivered impressive performance in various computer vision
applications. But the storage and computation requirements make it problematic
for deploying these models on mobile devices. Recently, tensor decompositions
have been used for speeding up CNNs. In this paper, we further develop the
tensor decomposition technique. We propose a new algorithm for computing the
low-rank tensor decomposition for removing the redundancy in the convolution
kernels. The algorithm finds the exact global optimizer of the decomposition
and is more effective than iterative methods. Based on the decomposition, we
further propose a new method for training low-rank constrained CNNs from
scratch. Interestingly, while achieving a significant speedup, sometimes the
low-rank constrained CNNs delivers significantly better performance than their
non-constrained counterparts. On the CIFAR-10 dataset, the proposed low-rank
NIN model achieves $91.31\%$ accuracy (without data augmentation), which also
improves upon state-of-the-art result. We evaluated the proposed method on
CIFAR-10 and ILSVRC12 datasets for a variety of modern CNNs, including AlexNet,
NIN, VGG and GoogleNet with success. For example, the forward time of VGG-16 is
reduced by half while the performance is still comparable. Empirical success
suggests that low-rank tensor decompositions can be a very useful tool for
speeding up large CNNs.
|
Reducing Overfitting in Deep Networks by Decorrelating Representations | One major challenge in training Deep Neural Networks is preventing
overfitting. Many techniques such as data augmentation and novel regularizers
such as Dropout have been proposed to prevent overfitting without requiring a
massive amount of training data. In this work, we propose a new regularizer
called DeCov which leads to significantly reduced overfitting (as indicated by
the difference between train and val performance), and better generalization.
Our regularizer encourages diverse or non-redundant representations in Deep
Neural Networks by minimizing the cross-covariance of hidden activations. This
simple intuition has been explored in a number of past works but surprisingly
has never been applied as a regularizer in supervised learning. Experiments
across a range of datasets and network architectures show that this loss always
reduces overfitting while almost always maintaining or increasing
generalization performance and often improving performance over Dropout.
|
Mediated Experts for Deep Convolutional Networks | We present a new supervised architecture termed Mediated Mixture-of-Experts
(MMoE) that allows us to improve classification accuracy of Deep Convolutional
Networks (DCN). Our architecture achieves this with the help of expert
networks: A network is trained on a disjoint subset of a given dataset and then
run in parallel to other experts during deployment. A mediator is employed if
experts contradict each other. This allows our framework to naturally support
incremental learning, as adding new classes requires (re-)training of the new
expert only. We also propose two measures to control computational complexity:
An early-stopping mechanism halts experts that have low confidence in their
prediction. The system allows to trade-off accuracy and complexity without
further retraining. We also suggest to share low-level convolutional layers
between experts in an effort to avoid computation of a near-duplicate feature
set. We evaluate our system on a popular dataset and report improved accuracy
compared to a single model of same configuration.
|
Learning Deep Structure-Preserving Image-Text Embeddings | This paper proposes a method for learning joint embeddings of images and text
using a two-branch neural network with multiple layers of linear projections
followed by nonlinearities. The network is trained using a large margin
objective that combines cross-view ranking constraints with within-view
neighborhood structure preservation constraints inspired by metric learning
literature. Extensive experiments show that our approach gains significant
improvements in accuracy for image-to-text and text-to-image retrieval. Our
method achieves new state-of-the-art results on the Flickr30K and MSCOCO
image-sentence datasets and shows promise on the new task of phrase
localization on the Flickr30K Entities dataset.
|
Variable Rate Image Compression with Recurrent Neural Networks | A large fraction of Internet traffic is now driven by requests from mobile
devices with relatively small screens and often stringent bandwidth
requirements. Due to these factors, it has become the norm for modern
graphics-heavy websites to transmit low-resolution, low-bytecount image
previews (thumbnails) as part of the initial page load process to improve
apparent page responsiveness. Increasing thumbnail compression beyond the
capabilities of existing codecs is therefore a current research focus, as any
byte savings will significantly enhance the experience of mobile device users.
Toward this end, we propose a general framework for variable-rate image
compression and a novel architecture based on convolutional and deconvolutional
LSTM recurrent networks. Our models address the main issues that have prevented
autoencoder neural networks from competing with existing image compression
algorithms: (1) our networks only need to be trained once (not per-image),
regardless of input image dimensions and the desired compression rate; (2) our
networks are progressive, meaning that the more bits are sent, the more
accurate the image reconstruction; and (3) the proposed architecture is at
least as efficient as a standard purpose-trained autoencoder for a given number
of bits. On a large-scale benchmark of 32$\times$32 thumbnails, our LSTM-based
approaches provide better visual quality than (headerless) JPEG, JPEG2000 and
WebP, with a storage size that is reduced by 10% or more.
|
Principled Parallel Mean-Field Inference for Discrete Random Fields | Mean-field variational inference is one of the most popular approaches to
inference in discrete random fields. Standard mean-field optimization is based
on coordinate descent and in many situations can be impractical. Thus, in
practice, various parallel techniques are used, which either rely on ad-hoc
smoothing with heuristically set parameters, or put strong constraints on the
type of models. In this paper, we propose a novel proximal gradient-based
approach to optimizing the variational objective. It is naturally
parallelizable and easy to implement. We prove its convergence, and then
demonstrate that, in practice, it yields faster convergence and often finds
better optima than more traditional mean-field optimization techniques.
Moreover, our method is less sensitive to the choice of parameters.
|
Semi-supervised Learning for Convolutional Neural Networks via Online
Graph Construction | The recent promising achievements of deep learning rely on the large amount
of labeled data. Considering the abundance of data on the web, most of them do
not have labels at all. Therefore, it is important to improve generalization
performance using unlabeled data on supervised tasks with few labeled
instances. In this work, we revisit graph-based semi-supervised learning
algorithms and propose an online graph construction technique which suits deep
convolutional neural network better. We consider an EM-like algorithm for
semi-supervised learning on deep neural networks: In forward pass, the graph is
constructed based on the network output, and the graph is then used for loss
calculation to help update the network by back propagation in the backward
pass. We demonstrate the strength of our online approach compared to the
conventional ones whose graph is constructed on static but not robust enough
feature representations beforehand.
|
Multi-task Sequence to Sequence Learning | Sequence to sequence learning has recently emerged as a new paradigm in
supervised learning. To date, most of its applications focused on only one task
and not much work explored this framework for multiple tasks. This paper
examines three multi-task learning (MTL) settings for sequence to sequence
models: (a) the oneto-many setting - where the encoder is shared between
several tasks such as machine translation and syntactic parsing, (b) the
many-to-one setting - useful when only the decoder can be shared, as in the
case of translation and image caption generation, and (c) the many-to-many
setting - where multiple encoders and decoders are shared, which is the case
with unsupervised objectives and translation. Our results show that training on
a small amount of parsing and image caption data can improve the translation
quality between English and German by up to 1.5 BLEU points over strong
single-task baselines on the WMT benchmarks. Furthermore, we have established a
new state-of-the-art result in constituent parsing with 93.0 F1. Lastly, we
reveal interesting properties of the two unsupervised learning objectives,
autoencoder and skip-thought, in the MTL context: autoencoder helps less in
terms of perplexities but more on BLEU scores compared to skip-thought.
|
Coreset-Based Adaptive Tracking | We propose a method for learning from streaming visual data using a compact,
constant size representation of all the data that was seen until a given
moment. Specifically, we construct a 'coreset' representation of streaming data
using a parallelized algorithm, which is an approximation of a set with
relation to the squared distances between this set and all other points in its
ambient space. We learn an adaptive object appearance model from the coreset
tree in constant time and logarithmic space and use it for object tracking by
detection. Our method obtains excellent results for object tracking on three
standard datasets over more than 100 videos. The ability to summarize data
efficiently makes our method ideally suited for tracking in long videos in
presence of space and time constraints. We demonstrate this ability by
outperforming a variety of algorithms on the TLD dataset with 2685 frames on
average. This coreset based learning approach can be applied for both real-time
learning of small, varied data and fast learning of big data.
|
Adjustable Bounded Rectifiers: Towards Deep Binary Representations | Binary representation is desirable for its memory efficiency, computation
speed and robustness. In this paper, we propose adjustable bounded rectifiers
to learn binary representations for deep neural networks. While hard
constraining representations across layers to be binary makes training
unreasonably difficult, we softly encourage activations to diverge from real
values to binary by approximating step functions. Our final representation is
completely binary. We test our approach on MNIST, CIFAR10, and ILSVRC2012
dataset, and systematically study the training dynamics of the binarization
process. Our approach can binarize the last layer representation without loss
of performance and binarize all the layers with reasonably small degradations.
The memory space that it saves may allow more sophisticated models to be
deployed, thus compensating the loss. To the best of our knowledge, this is the
first work to report results on current deep network architectures using
complete binary middle representations. Given the learned representations, we
find that the firing or inhibition of a binary neuron is usually associated
with a meaningful interpretation across different classes. This suggests that
the semantic structure of a neural network may be manifested through a guided
binarization process.
|
Diffusion Representations | Diffusion Maps framework is a kernel based method for manifold learning and
data analysis that defines diffusion similarities by imposing a Markovian
process on the given dataset. Analysis by this process uncovers the intrinsic
geometric structures in the data. Recently, it was suggested to replace the
standard kernel by a measure-based kernel that incorporates information about
the density of the data. Thus, the manifold assumption is replaced by a more
general measure-based assumption.
The measure-based diffusion kernel incorporates two separate independent
representations. The first determines a measure that correlates with a density
that represents normal behaviors and patterns in the data. The second consists
of the analyzed multidimensional data points.
In this paper, we present a representation framework for data analysis of
datasets that is based on a closed-form decomposition of the measure-based
kernel. The proposed representation preserves pairwise diffusion distances that
does not depend on the data size while being invariant to scale. For a
stationary data, no out-of-sample extension is needed for embedding newly
arrived data points in the representation space. Several aspects of the
presented methodology are demonstrated on analytically generated data.
|
Knowledge Base Population using Semantic Label Propagation | A crucial aspect of a knowledge base population system that extracts new
facts from text corpora, is the generation of training data for its relation
extractors. In this paper, we present a method that maximizes the effectiveness
of newly trained relation extractors at a minimal annotation cost. Manual
labeling can be significantly reduced by Distant Supervision, which is a method
to construct training data automatically by aligning a large text corpus with
an existing knowledge base of known facts. For example, all sentences
mentioning both 'Barack Obama' and 'US' may serve as positive training
instances for the relation born_in(subject,object). However, distant
supervision typically results in a highly noisy training set: many training
sentences do not really express the intended relation. We propose to combine
distant supervision with minimal manual supervision in a technique called
feature labeling, to eliminate noise from the large and noisy initial training
set, resulting in a significant increase of precision. We further improve on
this approach by introducing the Semantic Label Propagation method, which uses
the similarity between low-dimensional representations of candidate training
instances, to extend the training set in order to increase recall while
maintaining high precision. Our proposed strategy for generating training data
is studied and evaluated on an established test collection designed for
knowledge base population tasks. The experimental results show that the
Semantic Label Propagation strategy leads to substantial performance gains when
compared to existing approaches, while requiring an almost negligible manual
annotation effort.
|
Towards Open Set Deep Networks | Deep networks have produced significant gains for various visual recognition
problems, leading to high impact academic and commercial applications. Recent
work in deep networks highlighted that it is easy to generate images that
humans would never classify as a particular object class, yet networks classify
such images high confidence as that given class - deep network are easily
fooled with images humans do not consider meaningful. The closed set nature of
deep networks forces them to choose from one of the known classes leading to
such artifacts. Recognition in the real world is open set, i.e. the recognition
system should reject unknown/unseen classes at test time. We present a
methodology to adapt deep networks for open set recognition, by introducing a
new model layer, OpenMax, which estimates the probability of an input being
from an unknown class. A key element of estimating the unknown probability is
adapting Meta-Recognition concepts to the activation patterns in the
penultimate layer of the network. OpenMax allows rejection of "fooling" and
unrelated open set images presented to the system; OpenMax greatly reduces the
number of obvious errors made by a deep network. We prove that the OpenMax
concept provides bounded open space risk, thereby formally providing an open
set recognition solution. We evaluate the resulting open set deep networks
using pre-trained networks from the Caffe Model-zoo on ImageNet 2012 validation
data, and thousands of fooling and open set images. The proposed OpenMax model
significantly outperforms open set recognition accuracy of basic deep networks
as well as deep networks with thresholding of SoftMax probabilities.
|
Multimodal sparse representation learning and applications | Unsupervised methods have proven effective for discriminative tasks in a
single-modality scenario. In this paper, we present a multimodal framework for
learning sparse representations that can capture semantic correlation between
modalities. The framework can model relationships at a higher level by forcing
the shared sparse representation. In particular, we propose the use of joint
dictionary learning technique for sparse coding and formulate the joint
representation for concision, cross-modal representations (in case of a missing
modality), and union of the cross-modal representations. Given the accelerated
growth of multimodal data posted on the Web such as YouTube, Wikipedia, and
Twitter, learning good multimodal features is becoming increasingly important.
We show that the shared representations enabled by our framework substantially
improve the classification performance under both unimodal and multimodal
settings. We further show how deep architectures built on the proposed
framework are effective for the case of highly nonlinear correlations between
modalities. The effectiveness of our approach is demonstrated experimentally in
image denoising, multimedia event detection and retrieval on the TRECVID
dataset (audio-video), category classification on the Wikipedia dataset
(image-text), and sentiment classification on PhotoTweet (image-text).
|
Convolutional Clustering for Unsupervised Learning | The task of labeling data for training deep neural networks is daunting and
tedious, requiring millions of labels to achieve the current state-of-the-art
results. Such reliance on large amounts of labeled data can be relaxed by
exploiting hierarchical features via unsupervised learning techniques. In this
work, we propose to train a deep convolutional network based on an enhanced
version of the k-means clustering algorithm, which reduces the number of
correlated parameters in the form of similar filters, and thus increases test
categorization accuracy. We call our algorithm convolutional k-means
clustering. We further show that learning the connection between the layers of
a deep convolutional neural network improves its ability to be trained on a
smaller amount of labeled data. Our experiments show that the proposed
algorithm outperforms other techniques that learn filters unsupervised.
Specifically, we obtained a test accuracy of 74.1% on STL-10 and a test error
of 0.5% on MNIST.
|
Predicting online user behaviour using deep learning algorithms | We propose a robust classifier to predict buying intentions based on user
behaviour within a large e-commerce website. In this work we compare
traditional machine learning techniques with the most advanced deep learning
approaches. We show that both Deep Belief Networks and Stacked Denoising
auto-Encoders achieved a substantial improvement by extracting features from
high dimensional data during the pre-train phase. They prove also to be more
convenient to deal with severe class imbalance.
|
Stochastic modified equations and adaptive stochastic gradient
algorithms | We develop the method of stochastic modified equations (SME), in which
stochastic gradient algorithms are approximated in the weak sense by
continuous-time stochastic differential equations. We exploit the continuous
formulation together with optimal control theory to derive novel adaptive
hyper-parameter adjustment policies. Our algorithms have competitive
performance with the added benefit of being robust to varying models and
datasets. This provides a general methodology for the analysis and design of
stochastic gradient algorithms.
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.