title
stringlengths 7
246
| abstract
stringlengths 6
3.31k
|
|---|---|
Bandit Online Optimization Over the Permutahedron | The permutahedron is the convex polytope with vertex set consisting of the
vectors $(\pi(1),\dots, \pi(n))$ for all permutations (bijections) $\pi$ over
$\{1,\dots, n\}$. We study a bandit game in which, at each step $t$, an
adversary chooses a hidden weight weight vector $s_t$, a player chooses a
vertex $\pi_t$ of the permutahedron and suffers an observed loss of
$\sum_{i=1}^n \pi(i) s_t(i)$.
A previous algorithm CombBand of Cesa-Bianchi et al (2009) guarantees a
regret of $O(n\sqrt{T \log n})$ for a time horizon of $T$. Unfortunately,
CombBand requires at each step an $n$-by-$n$ matrix permanent approximation to
within improved accuracy as $T$ grows, resulting in a total running time that
is super linear in $T$, making it impractical for large time horizons.
We provide an algorithm of regret $O(n^{3/2}\sqrt{T})$ with total time
complexity $O(n^3T)$. The ideas are a combination of CombBand and a recent
algorithm by Ailon (2013) for online optimization over the permutahedron in the
full information setting. The technical core is a bound on the variance of the
Plackett-Luce noisy sorting process's "pseudo loss". The bound is obtained by
establishing positive semi-definiteness of a family of 3-by-3 matrices
generated from rational functions of exponentials of 3 parameters.
|
Max-Min Distance Nonnegative Matrix Factorization | Nonnegative Matrix Factorization (NMF) has been a popular representation
method for pattern classification problem. It tries to decompose a nonnegative
matrix of data samples as the product of a nonnegative basic matrix and a
nonnegative coefficient matrix, and the coefficient matrix is used as the new
representation. However, traditional NMF methods ignore the class labels of the
data samples. In this paper, we proposed a supervised novel NMF algorithm to
improve the discriminative ability of the new representation. Using the class
labels, we separate all the data sample pairs into within-class pairs and
between-class pairs. To improve the discriminate ability of the new NMF
representations, we hope that the maximum distance of the within-class pairs in
the new NMF space could be minimized, while the minimum distance of the
between-class pairs pairs could be maximized. With this criterion, we construct
an objective function and optimize it with regard to basic and coefficient
matrices and slack variables alternatively, resulting in a iterative algorithm.
|
Semi-Stochastic Gradient Descent Methods | In this paper we study the problem of minimizing the average of a large
number ($n$) of smooth convex loss functions. We propose a new method, S2GD
(Semi-Stochastic Gradient Descent), which runs for one or several epochs in
each of which a single full gradient and a random number of stochastic
gradients is computed, following a geometric law. The total work needed for the
method to output an $\varepsilon$-accurate solution in expectation, measured in
the number of passes over data, or equivalently, in units equivalent to the
computation of a single gradient of the loss, is
$O((\kappa/n)\log(1/\varepsilon))$, where $\kappa$ is the condition number.
This is achieved by running the method for $O(\log(1/\varepsilon))$ epochs,
with a single gradient evaluation and $O(\kappa)$ stochastic gradient
evaluations in each. The SVRG method of Johnson and Zhang arises as a special
case. If our method is limited to a single epoch only, it needs to evaluate at
most $O((\kappa/\varepsilon)\log(1/\varepsilon))$ stochastic gradients. In
contrast, SVRG requires $O(\kappa/\varepsilon^2)$ stochastic gradients. To
illustrate our theoretical results, S2GD only needs the workload equivalent to
about 2.1 full gradient evaluations to find an $10^{-6}$-accurate solution for
a problem with $n=10^9$ and $\kappa=10^3$.
|
Curriculum Learning for Handwritten Text Line Recognition | Recurrent Neural Networks (RNN) have recently achieved the best performance
in off-line Handwriting Text Recognition. At the same time, learning RNN by
gradient descent leads to slow convergence, and training times are particularly
long when the training database consists of full lines of text. In this paper,
we propose an easy way to accelerate stochastic gradient descent in this
set-up, and in the general context of learning to recognize sequences. The
principle is called Curriculum Learning, or shaping. The idea is to first learn
to recognize short sequences before training on all available training
sequences. Experiments on three different handwritten text databases (Rimes,
IAM, OpenHaRT) show that a simple implementation of this strategy can
significantly speed up the training of RNN for Text Recognition, and even
significantly improve performance in some cases.
|
Dual coordinate solvers for large-scale structural SVMs | This manuscript describes a method for training linear SVMs (including binary
SVMs, SVM regression, and structural SVMs) from large, out-of-core training
datasets. Current strategies for large-scale learning fall into one of two
camps; batch algorithms which solve the learning problem given a finite
datasets, and online algorithms which can process out-of-core datasets. The
former typically requires datasets small enough to fit in memory. The latter is
often phrased as a stochastic optimization problem; such algorithms enjoy
strong theoretical properties but often require manual tuned annealing
schedules, and may converge slowly for problems with large output spaces (e.g.,
structural SVMs). We discuss an algorithm for an "intermediate" regime in which
the data is too large to fit in memory, but the active constraints (support
vectors) are small enough to remain in memory. In this case, one can design
rather efficient learning algorithms that are as stable as batch algorithms,
but capable of processing out-of-core datasets. We have developed such a
MATLAB-based solver and used it to train a collection of recognition systems
for articulated pose estimation, facial analysis, 3D object recognition, and
action classification, all with publicly-available code. This writeup describes
the solver in detail.
|
Understanding Deep Architectures using a Recursive Convolutional Network | A key challenge in designing convolutional network models is sizing them
appropriately. Many factors are involved in these decisions, including number
of layers, feature maps, kernel sizes, etc. Complicating this further is the
fact that each of these influence not only the numbers and dimensions of the
activation units, but also the total number of parameters. In this paper we
focus on assessing the independent contributions of three of these linked
variables: The numbers of layers, feature maps, and parameters. To accomplish
this, we employ a recursive convolutional network whose weights are tied
between layers; this allows us to vary each of the three factors in a
controlled setting. We find that while increasing the numbers of layers and
parameters each have clear benefit, the number of feature maps (and hence
dimensionality of the representation) appears ancillary, and finds most of its
benefit through the introduction of more weights. Our results (i) empirically
confirm the notion that adding layers alone increases computational power,
within the context of convolutional layers, and (ii) suggest that precise
sizing of convolutional feature map dimensions is itself of little concern;
more attention should be paid to the number of parameters in these layers
instead.
|
From Maxout to Channel-Out: Encoding Information on Sparse Pathways | Motivated by an important insight from neural science, we propose a new
framework for understanding the success of the recently proposed "maxout"
networks. The framework is based on encoding information on sparse pathways and
recognizing the correct pathway at inference time. Elaborating further on this
insight, we propose a novel deep network architecture, called "channel-out"
network, which takes a much better advantage of sparse pathway encoding. In
channel-out networks, pathways are not only formed a posteriori, but they are
also actively selected according to the inference outputs from the lower
layers. From a mathematical perspective, channel-out networks can represent a
wider class of piece-wise continuous functions, thereby endowing the network
with more expressive power than that of maxout networks. We test our
channel-out networks on several well-known image classification benchmarks,
setting new state-of-the-art performance on CIFAR-100 and STL-10, which
represent some of the "harder" image classification benchmarks.
|
Robust Subspace System Identification via Weighted Nuclear Norm
Optimization | Subspace identification is a classical and very well studied problem in
system identification. The problem was recently posed as a convex optimization
problem via the nuclear norm relaxation. Inspired by robust PCA, we extend this
framework to handle outliers. The proposed framework takes the form of a convex
optimization problem with an objective that trades off fit, rank and sparsity.
As in robust PCA, it can be problematic to find a suitable regularization
parameter. We show how the space in which a suitable parameter should be sought
can be limited to a bounded open set of the two dimensional parameter space. In
practice, this is very useful since it restricts the parameter space that is
needed to be surveyed.
|
End-to-end Phoneme Sequence Recognition using Convolutional Neural
Networks | Most phoneme recognition state-of-the-art systems rely on a classical neural
network classifiers, fed with highly tuned features, such as MFCC or PLP
features. Recent advances in ``deep learning'' approaches questioned such
systems, but while some attempts were made with simpler features such as
spectrograms, state-of-the-art systems still rely on MFCCs. This might be
viewed as a kind of failure from deep learning approaches, which are often
claimed to have the ability to train with raw signals, alleviating the need of
hand-crafted features. In this paper, we investigate a convolutional neural
network approach for raw speech signals. While convolutional architectures got
tremendous success in computer vision or text processing, they seem to have
been let down in the past recent years in the speech processing field. We show
that it is possible to learn an end-to-end phoneme sequence classifier system
directly from raw signal, with similar performance on the TIMIT and WSJ
datasets than existing systems based on MFCC, questioning the need of complex
hand-crafted features on large datasets.
|
Sequential Monte Carlo Inference of Mixed Membership Stochastic
Blockmodels for Dynamic Social Networks | Many kinds of data can be represented as a network or graph. It is crucial to
infer the latent structure underlying such a network and to predict unobserved
links in the network. Mixed Membership Stochastic Blockmodel (MMSB) is a
promising model for network data. Latent variables and unknown parameters in
MMSB have been estimated through Bayesian inference with the entire network;
however, it is important to estimate them online for evolving networks. In this
paper, we first develop online inference methods for MMSB through sequential
Monte Carlo methods, also known as particle filters. We then extend them for
time-evolving networks, taking into account the temporal dependency of the
network structure. We demonstrate through experiments that the time-dependent
particle filter outperformed several baselines in terms of prediction
performance in an online condition.
|
Budgeted Influence Maximization for Multiple Products | The typical algorithmic problem in viral marketing aims to identify a set of
influential users in a social network, who, when convinced to adopt a product,
shall influence other users in the network and trigger a large cascade of
adoptions. However, the host (the owner of an online social platform) often
faces more constraints than a single product, endless user attentions,
unlimited budget and unbounded time; in reality, multiple products need to be
advertised, each user can tolerate only a small number of recommendations,
influencing user has a cost and advertisers have only limited budgets, and the
adoptions need to be maximized within a short time window.
Given theses myriads of user, monetary, and timing constraints, it is
extremely challenging for the host to design principled and efficient viral
market algorithms with provable guarantees. In this paper, we provide a novel
solution by formulating the problem as a submodular maximization in a
continuous-time diffusion model under an intersection of a matroid and multiple
knapsack constraints. We also propose an adaptive threshold greedy algorithm
which can be faster than the traditional greedy algorithm with lazy evaluation,
and scalable to networks with million of nodes. Furthermore, our mathematical
formulation allows us to prove that the algorithm can achieve an approximation
factor of $k_a/(2+2 k)$ when $k_a$ out of the $k$ knapsack constraints are
active, which also improves over previous guarantees from combinatorial
optimization literature. In the case when influencing each user has uniform
cost, the approximation becomes even better to a factor of $1/3$. Extensive
synthetic and real world experiments demonstrate that our budgeted influence
maximization algorithm achieves the-state-of-the-art in terms of both
effectiveness and scalability, often beating the next best by significant
margins.
|
bartMachine: Machine Learning with Bayesian Additive Regression Trees | We present a new package in R implementing Bayesian additive regression trees
(BART). The package introduces many new features for data analysis using BART
such as variable selection, interaction detection, model diagnostic plots,
incorporation of missing data and the ability to save trees for future
prediction. It is significantly faster than the current R implementation,
parallelized, and capable of handling both large sample sizes and
high-dimensional data.
|
Machine Learning Techniques for Intrusion Detection | An Intrusion Detection System (IDS) is a software that monitors a single or a
network of computers for malicious activities (attacks) that are aimed at
stealing or censoring information or corrupting network protocols. Most
techniques used in today's IDS are not able to deal with the dynamic and
complex nature of cyber attacks on computer networks. Hence, efficient adaptive
methods like various techniques of machine learning can result in higher
detection rates, lower false alarm rates and reasonable computation and
communication costs. In this paper, we study several such schemes and compare
their performance. We divide the schemes into methods based on classical
artificial intelligence (AI) and methods based on computational intelligence
(CI). We explain how various characteristics of CI techniques can be used to
build efficient IDS.
|
CEAI: CCM based Email Authorship Identification Model | In this paper we present a model for email authorship identification (EAI) by
employing a Cluster-based Classification (CCM) technique. Traditionally,
stylometric features have been successfully employed in various authorship
analysis tasks; we extend the traditional feature-set to include some more
interesting and effective features for email authorship identification (e.g.
the last punctuation mark used in an email, the tendency of an author to use
capitalization at the start of an email, or the punctuation after a greeting or
farewell). We also included Info Gain feature selection based content features.
It is observed that the use of such features in the authorship identification
process has a positive impact on the accuracy of the authorship identification
task. We performed experiments to justify our arguments and compared the
results with other base line models. Experimental results reveal that the
proposed CCM-based email authorship identification model, along with the
proposed feature set, outperforms the state-of-the-art support vector machine
(SVM)-based models, as well as the models proposed by Iqbal et al. [1, 2]. The
proposed model attains an accuracy rate of 94% for 10 authors, 89% for 25
authors, and 81% for 50 authors, respectively on Enron dataset, while 89.5%
accuracy has been achieved on authors' constructed real email dataset. The
results on Enron dataset have been achieved on quite a large number of authors
as compared to the models proposed by Iqbal et al. [1, 2].
|
Automatic recognition and tagging of topologically different regimes in
dynamical systems | Complex systems are commonly modeled using nonlinear dynamical systems. These
models are often high-dimensional and chaotic. An important goal in studying
physical systems through the lens of mathematical models is to determine when
the system undergoes changes in qualitative behavior. A detailed description of
the dynamics can be difficult or impossible to obtain for high-dimensional and
chaotic systems. Therefore, a more sensible goal is to recognize and mark
transitions of a system between qualitatively different regimes of behavior. In
practice, one is interested in developing techniques for detection of such
transitions from sparse observations, possibly contaminated by noise. In this
paper we develop a framework to accurately tag different regimes of complex
systems based on topological features. In particular, our framework works with
a high degree of success in picking out a cyclically orbiting regime from a
stationary equilibrium regime in high-dimensional stochastic dynamical systems.
|
Kernel-based Distance Metric Learning in the Output Space | In this paper we present two related, kernel-based Distance Metric Learning
(DML) methods. Their respective models non-linearly map data from their
original space to an output space, and subsequent distance measurements are
performed in the output space via a Mahalanobis metric. The dimensionality of
the output space can be directly controlled to facilitate the learning of a
low-rank metric. Both methods allow for simultaneous inference of the
associated metric and the mapping to the output space, which can be used to
visualize the data, when the output space is 2- or 3-dimensional. Experimental
results for a collection of classification tasks illustrate the advantages of
the proposed methods over other traditional and kernel-based DML approaches.
|
Multi-Task Classification Hypothesis Space with Improved Generalization
Bounds | This paper presents a RKHS, in general, of vector-valued functions intended
to be used as hypothesis space for multi-task classification. It extends
similar hypothesis spaces that have previously considered in the literature.
Assuming this space, an improved Empirical Rademacher Complexity-based
generalization bound is derived. The analysis is itself extended to an MKL
setting. The connection between the proposed hypothesis space and a Group-Lasso
type regularizer is discussed. Finally, experimental results, with some
SVM-based Multi-Task Learning problems, underline the quality of the derived
bounds and validate the paper's analysis.
|
Improving circuit miniaturization and its efficiency using Rough Set
Theory | High-speed, accuracy, meticulousness and quick response are notion of the
vital necessities for modern digital world. An efficient electronic circuit
unswervingly affects the maneuver of the whole system. Different tools are
required to unravel different types of engineering tribulations. Improving the
efficiency, accuracy and low power consumption in an electronic circuit is
always been a bottle neck problem. So the need of circuit miniaturization is
always there. It saves a lot of time and power that is wasted in switching of
gates, the wiring-crises is reduced, cross-sectional area of chip is reduced,
the number of transistors that can implemented in chip is multiplied many
folds. Therefore to trounce with this problem we have proposed an Artificial
intelligence (AI) based approach that make use of Rough Set Theory for its
implementation. Theory of rough set has been proposed by Z Pawlak in the year
1982. Rough set theory is a new mathematical tool which deals with uncertainty
and vagueness. Decisions can be generated using rough set theory by reducing
the unwanted and superfluous data. We have condensed the number of gates
without upsetting the productivity of the given circuit. This paper proposes an
approach with the help of rough set theory which basically lessens the number
of gates in the circuit, based on decision rules.
|
Performance Analysis Of Regularized Linear Regression Models For
Oxazolines And Oxazoles Derivitive Descriptor Dataset | Regularized regression techniques for linear regression have been created the
last few ten years to reduce the flaws of ordinary least squares regression
with regard to prediction accuracy. In this paper, new methods for using
regularized regression in model choice are introduced, and we distinguish the
conditions in which regularized regression develops our ability to discriminate
models. We applied all the five methods that use penalty-based (regularization)
shrinkage to handle Oxazolines and Oxazoles derivatives descriptor dataset with
far more predictors than observations. The lasso, ridge, elasticnet, lars and
relaxed lasso further possess the desirable property that they simultaneously
select relevant predictive descriptors and optimally estimate their effects.
Here, we comparatively evaluate the performance of five regularized linear
regression methods The assessment of the performance of each model by means of
benchmark experiments is an established exercise. Cross-validation and
resampling methods are generally used to arrive point evaluates the
efficiencies which are compared to recognize methods with acceptable features.
Predictive accuracy was evaluated using the root mean squared error (RMSE) and
Square of usual correlation between predictors and observed mean inhibitory
concentration of antitubercular activity (R square). We found that all five
regularized regression models were able to produce feasible models and
efficient capturing the linearity in the data. The elastic net and lars had
similar accuracies as well as lasso and relaxed lasso had similar accuracies
but outperformed ridge regression in terms of the RMSE and R square metrics.
|
Active Player Modelling | We argue for the use of active learning methods for player modelling. In
active learning, the learning algorithm chooses where to sample the search
space so as to optimise learning progress. We hypothesise that player modelling
based on active learning could result in vastly more efficient learning, but
will require big changes in how data is collected. Some example active player
modelling scenarios are described. A particular form of active learning is also
equivalent to an influential formalisation of (human and machine) curiosity,
and games with active learning could therefore be seen as being curious about
the player. We further hypothesise that this form of curiosity is symmetric,
and therefore that games that explore their players based on the principles of
active learning will turn out to select game configurations that are
interesting to the player that is being explored.
|
Protein Contact Prediction by Integrating Joint Evolutionary Coupling
Analysis and Supervised Learning | Protein contacts contain important information for protein structure and
functional study, but contact prediction from sequence remains very
challenging. Both evolutionary coupling (EC) analysis and supervised machine
learning methods are developed to predict contacts, making use of different
types of information, respectively. This paper presents a group graphical lasso
(GGL) method for contact prediction that integrates joint multi-family EC
analysis and supervised learning. Different from existing single-family EC
analysis that uses residue co-evolution information in only the target protein
family, our joint EC analysis uses residue co-evolution in both the target
family and its related families, which may have divergent sequences but similar
folds. To implement joint EC analysis, we model a set of related protein
families using Gaussian graphical models (GGM) and then co-estimate their
precision matrices by maximum-likelihood, subject to the constraint that the
precision matrices shall share similar residue co-evolution patterns. To
further improve the accuracy of the estimated precision matrices, we employ a
supervised learning method to predict contact probability from a variety of
evolutionary and non-evolutionary information and then incorporate the
predicted probability as prior into our GGL framework. Experiments show that
our method can predict contacts much more accurately than existing methods, and
that our method performs better on both conserved and family-specific contacts.
|
Clustering for high-dimension, low-sample size data using distance
vectors | In high-dimension, low-sample size (HDLSS) data, it is not always true that
closeness of two objects reflects a hidden cluster structure. We point out the
important fact that it is not the closeness, but the "values" of distance that
contain information of the cluster structure in high-dimensional space. Based
on this fact, we propose an efficient and simple clustering approach, called
distance vector clustering, for HDLSS data. Under the assumptions given in the
work of Hall et al. (2005), we show the proposed approach provides a true
cluster label under milder conditions when the dimension tends to infinity with
the sample size fixed. The effectiveness of the distance vector clustering
approach is illustrated through a numerical experiment and real data analysis.
|
Online Bayesian Passive-Aggressive Learning | Online Passive-Aggressive (PA) learning is an effective framework for
performing max-margin online learning. But the deterministic formulation and
estimated single large-margin model could limit its capability in discovering
descriptive structures underlying complex data. This pa- per presents online
Bayesian Passive-Aggressive (BayesPA) learning, which subsumes the online PA
and extends naturally to incorporate latent variables and perform nonparametric
Bayesian inference, thus providing great flexibility for explorative analysis.
We apply BayesPA to topic modeling and derive efficient online learning
algorithms for max-margin topic models. We further develop nonparametric
methods to resolve the number of topics. Experimental results on real datasets
show that our approaches significantly improve time efficiency while
maintaining comparable results with the batch counterparts.
|
Relative Upper Confidence Bound for the K-Armed Dueling Bandit Problem | This paper proposes a new method for the K-armed dueling bandit problem, a
variation on the regular K-armed bandit problem that offers only relative
feedback about pairs of arms. Our approach extends the Upper Confidence Bound
algorithm to the relative setting by using estimates of the pairwise
probabilities to select a promising arm and applying Upper Confidence Bound
with the winner as a benchmark. We prove a finite-time regret bound of order
O(log t). In addition, our empirical results using real data from an
information retrieval application show that it greatly outperforms the state of
the art.
|
Unsupervised learning of depth and motion | We present a model for the joint estimation of disparity and motion. The
model is based on learning about the interrelations between images from
multiple cameras, multiple frames in a video, or the combination of both. We
show that learning depth and motion cues, as well as their combinations, from
data is possible within a single type of architecture and a single type of
learning algorithm, by using biologically inspired "complex cell" like units,
which encode correlations between the pixels across image pairs. Our
experimental results show that the learning of depth and motion makes it
possible to achieve state-of-the-art performance in 3-D activity analysis, and
to outperform existing hand-engineered 3-D motion features by a very large
margin.
|
Sparse Matrix-based Random Projection for Classification | As a typical dimensionality reduction technique, random projection can be
simply implemented with linear projection, while maintaining the pairwise
distances of high-dimensional data with high probability. Considering this
technique is mainly exploited for the task of classification, this paper is
developed to study the construction of random matrix from the viewpoint of
feature selection, rather than of traditional distance preservation. This
yields a somewhat surprising theoretical result, that is, the sparse random
matrix with exactly one nonzero element per column, can present better feature
selection performance than other more dense matrices, if the projection
dimension is sufficiently large (namely, not much smaller than the number of
feature elements); otherwise, it will perform comparably to others. For random
projection, this theoretical result implies considerable improvement on both
complexity and performance, which is widely confirmed with the classification
experiments on both synthetic data and real data.
|
Efficient Baseline-free Sampling in Parameter Exploring Policy
Gradients: Super Symmetric PGPE | Policy Gradient methods that explore directly in parameter space are among
the most effective and robust direct policy search methods and have drawn a lot
of attention lately. The basic method from this field, Policy Gradients with
Parameter-based Exploration, uses two samples that are symmetric around the
current hypothesis to circumvent misleading reward in \emph{asymmetrical}
reward distributed problems gathered with the usual baseline approach. The
exploration parameters are still updated by a baseline approach - leaving the
exploration prone to asymmetric reward distributions. In this paper we will
show how the exploration parameters can be sampled quasi symmetric despite
having limited instead of free parameters for exploration. We give a
transformation approximation to get quasi symmetric samples with respect to the
exploration without changing the overall sampling distribution. Finally we will
demonstrate that sampling symmetrically also for the exploration parameters is
superior in needs of samples and robustness than the original sampling
approach.
|
A Methodology for Player Modeling based on Machine Learning | AI is gradually receiving more attention as a fundamental feature to increase
the immersion in digital games. Among the several AI approaches, player
modeling is becoming an important one. The main idea is to understand and model
the player characteristics and behaviors in order to develop a better AI. In
this work, we discuss several aspects of this new field. We proposed a taxonomy
to organize the area, discussing several facets of this topic, ranging from
implementation decisions up to what a model attempts to describe. We then
classify, in our taxonomy, some of the most important works in this field. We
also presented a generic approach to deal with player modeling using ML, and we
instantiated this approach to model players' preferences in the game
Civilization IV. The instantiation of this approach has several steps. We first
discuss a generic representation, regardless of what is being modeled, and
evaluate it performing experiments with the strategy game Civilization IV.
Continuing the instantiation of the proposed approach we evaluated the
applicability of using game score information to distinguish different
preferences. We presented a characterization of virtual agents in the game,
comparing their behavior with their stated preferences. Once we have
characterized these agents, we were able to observe that different preferences
generate different behaviors, measured by several game indicators. We then
tackled the preference modeling problem as a binary classification task, with a
supervised learning approach. We compared four different methods, based on
different paradigms (SVM, AdaBoost, NaiveBayes and JRip), evaluating them on a
set of matches played by different virtual agents. We conclude our work using
the learned models to infer human players' preferences. Using some of the
evaluated classifiers we obtained accuracies over 60% for most of the inferred
preferences.
|
An Extensive Evaluation of Filtering Misclassified Instances in
Supervised Classification Tasks | Removing or filtering outliers and mislabeled instances prior to training a
learning algorithm has been shown to increase classification accuracy. A
popular approach for handling outliers and mislabeled instances is to remove
any instance that is misclassified by a learning algorithm. However, an
examination of which learning algorithms to use for filtering as well as their
effects on multiple learning algorithms over a large set of data sets has not
been done. Previous work has generally been limited due to the large
computational requirements to run such an experiment, and, thus, the
examination has generally been limited to learning algorithms that are
computationally inexpensive and using a small number of data sets. In this
paper, we examine 9 learning algorithms as filtering algorithms as well as
examining the effects of filtering in the 9 chosen learning algorithms on a set
of 54 data sets. In addition to using each learning algorithm individually as a
filter, we also use the set of learning algorithms as an ensemble filter and
use an adaptive algorithm that selects a subset of the learning algorithms for
filtering for a specific task and learning algorithm. We find that for most
cases, using an ensemble of learning algorithms for filtering produces the
greatest increase in classification accuracy. We also compare filtering with a
majority voting ensemble. The voting ensemble significantly outperforms
filtering unless there are high amounts of noise present in the data set.
Additionally, we find that a majority voting ensemble is robust to noise as
filtering with a voting ensemble does not increase the classification accuracy
of the voting ensemble.
|
Classifiers With a Reject Option for Early Time-Series Classification | Early classification of time-series data in a dynamic environment is a
challenging problem of great importance in signal processing. This paper
proposes a classifier architecture with a reject option capable of online
decision making without the need to wait for the entire time series signal to
be present. The main idea is to classify an odor/gas signal with an acceptable
accuracy as early as possible. Instead of using posterior probability of a
classifier, the proposed method uses the "agreement" of an ensemble to decide
whether to accept or reject the candidate label. The introduced algorithm is
applied to the bio-chemistry problem of odor classification to build a novel
Electronic-Nose called Forefront-Nose. Experimental results on wind tunnel
test-bed facility confirms the robustness of the forefront-nose compared to the
standard classifiers from both earliness and recognition perspectives.
|
ECOC-Based Training of Neural Networks for Face Recognition | Error Correcting Output Codes, ECOC, is an output representation method
capable of discovering some of the errors produced in classification tasks.
This paper describes the application of ECOC to the training of feed forward
neural networks, FFNN, for improving the overall accuracy of classification
systems. Indeed, to improve the generalization of FFNN classifiers, this paper
proposes an ECOC-Based training method for Neural Networks that use ECOC as the
output representation, and adopts the traditional Back-Propagation algorithm,
BP, to adjust weights of the network. Experimental results for face recognition
problem on Yale database demonstrate the effectiveness of our method. With a
rejection scheme defined by a simple robustness rate, high reliability is
achieved in this application.
|
Domain adaptation for sequence labeling using hidden Markov models | Most natural language processing systems based on machine learning are not
robust to domain shift. For example, a state-of-the-art syntactic dependency
parser trained on Wall Street Journal sentences has an absolute drop in
performance of more than ten points when tested on textual data from the Web.
An efficient solution to make these methods more robust to domain shift is to
first learn a word representation using large amounts of unlabeled data from
both domains, and then use this representation as features in a supervised
learning algorithm. In this paper, we propose to use hidden Markov models to
learn word representations for part-of-speech tagging. In particular, we study
the influence of using data from the source, the target or both domains to
learn the representation and the different ways to represent words using an
HMM.
|
A MapReduce based distributed SVM algorithm for binary classification | Although Support Vector Machine (SVM) algorithm has a high generalization
property to classify for unseen examples after training phase and it has small
loss value, the algorithm is not suitable for real-life classification and
regression problems. SVMs cannot solve hundreds of thousands examples in
training dataset. In previous studies on distributed machine learning
algorithms, SVM is trained over a costly and preconfigured computer
environment. In this research, we present a MapReduce based distributed
parallel SVM training algorithm for binary classification problems. This work
shows how to distribute optimization problem over cloud computing systems with
MapReduce technique. In the second step of this work, we used statistical
learning theory to find the predictive hypothesis that minimize our empirical
risks from hypothesis spaces that created with reduce function of MapReduce.
The results of this research are important for training of big datasets for SVM
algorithm based classification problems. We provided that iterative training of
split dataset with MapReduce technique; accuracy of the classifier function
will converge to global optimal classifier function's accuracy in finite
iteration size. The algorithm performance was measured on samples from letter
recognition and pen-based recognition of handwritten digits dataset.
|
Distributed k-means algorithm | In this paper we provide a fully distributed implementation of the k-means
clustering algorithm, intended for wireless sensor networks where each agent is
endowed with a possibly high-dimensional observation (e.g., position, humidity,
temperature, etc.) The proposed algorithm, by means of one-hop communication,
partitions the agents into measure-dependent groups that have small in-group
and large out-group "distances". Since the partitions may not have a relation
with the topology of the network--members of the same clusters may not be
spatially close--the algorithm is provided with a mechanism to compute the
clusters'centroids even when the clusters are disconnected in several
sub-clusters.The results of the proposed distributed algorithm coincide, in
terms of minimization of the objective function, with the centralized k-means
algorithm. Some numerical examples illustrate the capabilities of the proposed
solution.
|
Feature Graph Architectures | In this article we propose feature graph architectures (FGA), which are deep
learning systems employing a structured initialisation and training method
based on a feature graph which facilitates improved generalisation performance
compared with a standard shallow architecture. The goal is to explore
alternative perspectives on the problem of deep network training. We evaluate
FGA performance for deep SVMs on some experimental datasets, and show how
generalisation and stability results may be derived for these models. We
describe the effect of permutations on the model accuracy, and give a criterion
for the optimal permutation in terms of feature correlations. The experimental
results show that the algorithm produces robust and significant test set
improvements over a standard shallow SVM training method for a range of
datasets. These gains are achieved with a moderate increase in time complexity.
|
Learning Factored Representations in a Deep Mixture of Experts | Mixtures of Experts combine the outputs of several "expert" networks, each of
which specializes in a different part of the input space. This is achieved by
training a "gating" network that maps each input to a distribution over the
experts. Such models show promise for building larger networks that are still
cheap to compute at test time, and more parallelizable at training time. In
this this work, we extend the Mixture of Experts to a stacked model, the Deep
Mixture of Experts, with multiple sets of gating and experts. This
exponentially increases the number of effective experts by associating each
input with a combination of experts at each layer, yet maintains a modest model
size. On a randomly translated version of the MNIST dataset, we find that the
Deep Mixture of Experts automatically learns to develop location-dependent
("where") experts at the first layer, and class-specific ("what") experts at
the second layer. In addition, we see that the different combinations are in
use when the model is applied to a dataset of speech monophones. These
demonstrate effective use of all expert combinations.
|
Rectifying Self Organizing Maps for Automatic Concept Learning from Web
Images | We attack the problem of learning concepts automatically from noisy web image
search results. Going beyond low level attributes, such as colour and texture,
we explore weakly-labelled datasets for the learning of higher level concepts,
such as scene categories. The idea is based on discovering common
characteristics shared among subsets of images by posing a method that is able
to organise the data while eliminating irrelevant instances. We propose a novel
clustering and outlier detection method, namely Rectifying Self Organizing Maps
(RSOM). Given an image collection returned for a concept query, RSOM provides
clusters pruned from outliers. Each cluster is used to train a model
representing a different characteristics of the concept. The proposed method
outperforms the state-of-the-art studies on the task of learning low-level
concepts, and it is competitive in learning higher level concepts as well. It
is capable to work at large scale with no supervision through exploiting the
available sources.
|
Network In Network | We propose a novel deep network structure called "Network In Network" (NIN)
to enhance model discriminability for local patches within the receptive field.
The conventional convolutional layer uses linear filters followed by a
nonlinear activation function to scan the input. Instead, we build micro neural
networks with more complex structures to abstract the data within the receptive
field. We instantiate the micro neural network with a multilayer perceptron,
which is a potent function approximator. The feature maps are obtained by
sliding the micro networks over the input in a similar manner as CNN; they are
then fed into the next layer. Deep NIN can be implemented by stacking mutiple
of the above described structure. With enhanced local modeling via the micro
network, we are able to utilize global average pooling over feature maps in the
classification layer, which is easier to interpret and less prone to
overfitting than traditional fully connected layers. We demonstrated the
state-of-the-art classification performances with NIN on CIFAR-10 and
CIFAR-100, and reasonable performances on SVHN and MNIST datasets.
|
Learning Deep Representations By Distributed Random Samplings | In this paper, we propose an extremely simple deep model for the unsupervised
nonlinear dimensionality reduction -- deep distributed random samplings, which
performs like a stack of unsupervised bootstrap aggregating. First, its network
structure is novel: each layer of the network is a group of mutually
independent $k$-centers clusterings. Second, its learning method is extremely
simple: the $k$ centers of each clustering are only $k$ randomly selected
examples from the training data; for small-scale data sets, the $k$ centers are
further randomly reconstructed by a simple cyclic-shift operation. Experimental
results on nonlinear dimensionality reduction show that the proposed method can
learn abstract representations on both large-scale and small-scale problems,
and meanwhile is much faster than deep neural networks on large-scale problems.
|
Optimization for Compressed Sensing: the Simplex Method and Kronecker
Sparsification | In this paper we present two new approaches to efficiently solve large-scale
compressed sensing problems. These two ideas are independent of each other and
can therefore be used either separately or together. We consider all
possibilities.
For the first approach, we note that the zero vector can be taken as the
initial basic (infeasible) solution for the linear programming problem and
therefore, if the true signal is very sparse, some variants of the simplex
method can be expected to take only a small number of pivots to arrive at a
solution. We implemented one such variant and demonstrate a dramatic
improvement in computation time on very sparse signals.
The second approach requires a redesigned sensing mechanism in which the
vector signal is stacked into a matrix. This allows us to exploit the Kronecker
compressed sensing (KCS) mechanism. We show that the Kronecker sensing requires
stronger conditions for perfect recovery compared to the original vector
problem. However, the Kronecker sensing, modeled correctly, is a much sparser
linear optimization problem. Hence, algorithms that benefit from sparse problem
representation, such as interior-point methods, can solve the Kronecker sensing
problems much faster than the corresponding vector problem. In our numerical
studies, we demonstrate a ten-fold improvement in the computation time.
|
Low-Rank Approximations for Conditional Feedforward Computation in Deep
Neural Networks | Scalability properties of deep neural networks raise key research questions,
particularly as the problems considered become larger and more challenging.
This paper expands on the idea of conditional computation introduced by Bengio,
et. al., where the nodes of a deep network are augmented by a set of gating
units that determine when a node should be calculated. By factorizing the
weight matrix into a low-rank approximation, an estimation of the sign of the
pre-nonlinearity activation can be efficiently obtained. For networks using
rectified-linear hidden units, this implies that the computation of a hidden
unit with an estimated negative pre-nonlinearity can be ommitted altogether, as
its value will become zero when nonlinearity is applied. For sparse neural
networks, this can result in considerable speed gains. Experimental results
using the MNIST and SVHN data sets with a fully-connected deep neural network
demonstrate the performance robustness of the proposed scheme with respect to
the error introduced by the conditional computation process.
|
Parametric Modelling of Multivariate Count Data Using Probabilistic
Graphical Models | Multivariate count data are defined as the number of items of different
categories issued from sampling within a population, which individuals are
grouped into categories. The analysis of multivariate count data is a recurrent
and crucial issue in numerous modelling problems, particularly in the fields of
biology and ecology (where the data can represent, for example, children counts
associated with multitype branching processes), sociology and econometrics. We
focus on I) Identifying categories that appear simultaneously, or on the
contrary that are mutually exclusive. This is achieved by identifying
conditional independence relationships between the variables; II)Building
parsimonious parametric models consistent with these relationships; III)
Characterising and testing the effects of covariates on the joint distribution
of the counts. To achieve these goals, we propose an approach based on
graphical probabilistic models, and more specifically partially directed
acyclic graphs.
|
Probable convexity and its application to Correlated Topic Models | Non-convex optimization problems often arise from probabilistic modeling,
such as estimation of posterior distributions. Non-convexity makes the problems
intractable, and poses various obstacles for us to design efficient algorithms.
In this work, we attack non-convexity by first introducing the concept of
\emph{probable convexity} for analyzing convexity of real functions in
practice. We then use the new concept to analyze an inference problem in the
\emph{Correlated Topic Model} (CTM) and related nonconjugate models. Contrary
to the existing belief of intractability, we show that this inference problem
is concave under certain conditions. One consequence of our analyses is a novel
algorithm for learning CTM which is significantly more scalable and qualitative
than existing methods. Finally, we highlight that stochastic gradient
algorithms might be a practical choice to resolve efficiently non-convex
problems. This finding might find beneficial in many contexts which are beyond
probabilistic modeling.
|
Comparative Analysis of Viterbi Training and Maximum Likelihood
Estimation for HMMs | We present an asymptotic analysis of Viterbi Training (VT) and contrast it
with a more conventional Maximum Likelihood (ML) approach to parameter
estimation in Hidden Markov Models. While ML estimator works by (locally)
maximizing the likelihood of the observed data, VT seeks to maximize the
probability of the most likely hidden state sequence. We develop an analytical
framework based on a generating function formalism and illustrate it on an
exactly solvable model of HMM with one unambiguous symbol. For this particular
model the ML objective function is continuously degenerate. VT objective, in
contrast, is shown to have only finite degeneracy. Furthermore, VT converges
faster and results in sparser (simpler) models, thus realizing an automatic
Occam's razor for HMM learning. For more general scenario VT can be worse
compared to ML but still capable of correctly recovering most of the
parameters.
|
Adaptive Stochastic Alternating Direction Method of Multipliers | The Alternating Direction Method of Multipliers (ADMM) has been studied for
years. The traditional ADMM algorithm needs to compute, at each iteration, an
(empirical) expected loss function on all training examples, resulting in a
computational complexity proportional to the number of training examples. To
reduce the time complexity, stochastic ADMM algorithms were proposed to replace
the expected function with a random loss function associated with one uniformly
drawn example plus a Bregman divergence. The Bregman divergence, however, is
derived from a simple second order proximal function, the half squared norm,
which could be a suboptimal choice.
In this paper, we present a new family of stochastic ADMM algorithms with
optimal second order proximal functions, which produce a new family of adaptive
subgradient methods. We theoretically prove that their regret bounds are as
good as the bounds which could be achieved by the best proximal function that
can be chosen in hindsight. Encouraging empirical results on a variety of
real-world datasets confirm the effectiveness and efficiency of the proposed
algorithms.
|
Dropout improves Recurrent Neural Networks for Handwriting Recognition | Recurrent neural networks (RNNs) with Long Short-Term memory cells currently
hold the best known results in unconstrained handwriting recognition. We show
that their performance can be greatly improved using dropout - a recently
proposed regularization method for deep architectures. While previous works
showed that dropout gave superior performance in the context of convolutional
networks, it had never been applied to RNNs. In our approach, dropout is
carefully used in the network so that it does not affect the recurrent
connections, hence the power of RNNs in modeling sequence is preserved.
Extensive experiments on a broad range of handwritten databases confirm the
effectiveness of dropout on deep architectures even when the network mainly
consists of recurrent and shared connections.
|
Evolution and Computational Learning Theory: A survey on Valiant's paper | Darwin's theory of evolution is considered to be one of the greatest
scientific gems in modern science. It not only gives us a description of how
living things evolve, but also shows how a population evolves through time and
also, why only the fittest individuals continue the generation forward. The
paper basically gives a high level analysis of the works of Valiant[1]. Though,
we know the mechanisms of evolution, but it seems that there does not exist any
strong quantitative and mathematical theory of the evolution of certain
mechanisms. What is defined exactly as the fitness of an individual, why is
that only certain individuals in a population tend to mutate, how computation
is done in finite time when we have exponentially many examples: there seems to
be a lot of questions which need to be answered. [1] basically treats Darwinian
theory as a form of computational learning theory, which calculates the net
fitness of the hypotheses and thus distinguishes functions and their classes
which could be evolvable using polynomial amount of resources. Evolution is
considered as a function of the environment and the previous evolutionary
stages that chooses the best hypothesis using learning techniques that makes
mutation possible and hence, gives a quantitative idea that why only the
fittest individuals tend to survive and have the power to mutate.
|
Compact Random Feature Maps | Kernel approximation using randomized feature maps has recently gained a lot
of interest. In this work, we identify that previous approaches for polynomial
kernel approximation create maps that are rank deficient, and therefore do not
utilize the capacity of the projected feature space effectively. To address
this challenge, we propose compact random feature maps (CRAFTMaps) to
approximate polynomial kernels more concisely and accurately. We prove the
error bounds of CRAFTMaps demonstrating their superior kernel reconstruction
performance compared to the previous approximation schemes. We show how
structured random matrices can be used to efficiently generate CRAFTMaps, and
present a single-pass algorithm using CRAFTMaps to learn non-linear multi-class
classifiers. We present experiments on multiple standard data-sets with
performance competitive with state-of-the-art results.
|
Sparse, complex-valued representations of natural sounds learned with
phase and amplitude continuity priors | Complex-valued sparse coding is a data representation which employs a
dictionary of two-dimensional subspaces, while imposing a sparse, factorial
prior on complex amplitudes. When trained on a dataset of natural image
patches, it learns phase invariant features which closely resemble receptive
fields of complex cells in the visual cortex. Features trained on natural
sounds however, rarely reveal phase invariance and capture other aspects of the
data. This observation is a starting point of the present work. As its first
contribution, it provides an analysis of natural sound statistics by means of
learning sparse, complex representations of short speech intervals. Secondly,
it proposes priors over the basis function set, which bias them towards
phase-invariant solutions. In this way, a dictionary of complex basis functions
can be learned from the data statistics, while preserving the phase invariance
property. Finally, representations trained on speech sounds with and without
priors are compared. Prior-based basis functions reveal performance comparable
to unconstrained sparse coding, while explicitely representing phase as a
temporal shift. Such representations can find applications in many perceptual
and machine learning tasks.
|
A Comparative Evaluation of Curriculum Learning with Filtering and
Boosting | Not all instances in a data set are equally beneficial for inferring a model
of the data. Some instances (such as outliers) are detrimental to inferring a
model of the data. Several machine learning techniques treat instances in a
data set differently during training such as curriculum learning, filtering,
and boosting. However, an automated method for determining how beneficial an
instance is for inferring a model of the data does not exist. In this paper, we
present an automated method that orders the instances in a data set by
complexity based on the their likelihood of being misclassified (instance
hardness). The underlying assumption of this method is that instances with a
high likelihood of being misclassified represent more complex concepts in a
data set. Ordering the instances in a data set allows a learning algorithm to
focus on the most beneficial instances and ignore the detrimental ones. We
compare ordering the instances in a data set in curriculum learning, filtering
and boosting. We find that ordering the instances significantly increases
classification accuracy and that filtering has the largest impact on
classification accuracy. On a set of 52 data sets, ordering the instances
increases the average accuracy from 81% to 84%.
|
Efficient Online Bootstrapping for Large Scale Learning | Bootstrapping is a useful technique for estimating the uncertainty of a
predictor, for example, confidence intervals for prediction. It is typically
used on small to moderate sized datasets, due to its high computation cost.
This work describes a highly scalable online bootstrapping strategy,
implemented inside Vowpal Wabbit, that is several times faster than traditional
strategies. Our experiments indicate that, in addition to providing a black
box-like method for estimating uncertainty, our implementation of online
bootstrapping may also help to train models with better prediction performance
due to model averaging.
|
Contextually Supervised Source Separation with Application to Energy
Disaggregation | We propose a new framework for single-channel source separation that lies
between the fully supervised and unsupervised setting. Instead of supervision,
we provide input features for each source signal and use convex methods to
estimate the correlations between these features and the unobserved signal
decomposition. We analyze the case of $\ell_2$ loss theoretically and show that
recovery of the signal components depends only on cross-correlation between
features for different signals, not on correlations between features for the
same signal. Contextually supervised source separation is a natural fit for
domains with large amounts of data but no explicit supervision; our motivating
application is energy disaggregation of hourly smart meter data (the separation
of whole-home power signals into different energy uses). Here we apply
contextual supervision to disaggregate the energy usage of thousands homes over
four years, a significantly larger scale than previously published efforts, and
demonstrate on synthetic data that our method outperforms the unsupervised
approach.
|
Permuted NMF: A Simple Algorithm Intended to Minimize the Volume of the
Score Matrix | Non-Negative Matrix Factorization, NMF, attempts to find a number of
archetypal response profiles, or parts, such that any sample profile in the
dataset can be approximated by a close profile among these archetypes or a
linear combination of these profiles. The non-negativity constraint is imposed
while estimating archetypal profiles, due to the non-negative nature of the
observed signal. Apart from non negativity, a volume constraint can be applied
on the Score matrix W to enhance the ability of learning parts of NMF. In this
report, we describe a very simple algorithm, which in effect achieves volume
minimization, although indirectly.
|
The Total Variation on Hypergraphs - Learning on Hypergraphs Revisited | Hypergraphs allow one to encode higher-order relationships in data and are
thus a very flexible modeling tool. Current learning methods are either based
on approximations of the hypergraphs via graphs or on tensor methods which are
only applicable under special conditions. In this paper, we present a new
learning framework on hypergraphs which fully uses the hypergraph structure.
The key element is a family of regularization functionals based on the total
variation on hypergraphs.
|
Nonlinear Eigenproblems in Data Analysis - Balanced Graph Cuts and the
RatioDCA-Prox | It has been recently shown that a large class of balanced graph cuts allows
for an exact relaxation into a nonlinear eigenproblem. We review briefly some
of these results and propose a family of algorithms to compute nonlinear
eigenvectors which encompasses previous work as special cases. We provide a
detailed analysis of the properties and the convergence behavior of these
algorithms and then discuss their application in the area of balanced graph
cuts.
|
Learning Semantic Script Knowledge with Event Embeddings | Induction of common sense knowledge about prototypical sequences of events
has recently received much attention. Instead of inducing this knowledge in the
form of graphs, as in much of the previous work, in our method, distributed
representations of event realizations are computed based on distributed
representations of predicates and their arguments, and then these
representations are used to predict prototypical event orderings. The
parameters of the compositional process for computing the event representations
and the ranking component of the model are jointly estimated from texts. We
show that this approach results in a substantial boost in ordering performance
with respect to previous methods.
|
Unsupervised feature learning by augmenting single images | When deep learning is applied to visual object recognition, data augmentation
is often used to generate additional training data without extra labeling cost.
It helps to reduce overfitting and increase the performance of the algorithm.
In this paper we investigate if it is possible to use data augmentation as the
main component of an unsupervised feature learning architecture. To that end we
sample a set of random image patches and declare each of them to be a separate
single-image surrogate class. We then extend these trivial one-element classes
by applying a variety of transformations to the initial 'seed' patches. Finally
we train a convolutional neural network to discriminate between these surrogate
classes. The feature representation learned by the network can then be used in
various vision tasks. We find that this simple feature learning algorithm is
surprisingly successful, achieving competitive classification results on
several popular vision datasets (STL-10, CIFAR-10, Caltech-101).
|
On the Challenges of Physical Implementations of RBMs | Restricted Boltzmann machines (RBMs) are powerful machine learning models,
but learning and some kinds of inference in the model require sampling-based
approximations, which, in classical digital computers, are implemented using
expensive MCMC. Physical computation offers the opportunity to reduce the cost
of sampling by building physical systems whose natural dynamics correspond to
drawing samples from the desired RBM distribution. Such a system avoids the
burn-in and mixing cost of a Markov chain. However, hardware implementations of
this variety usually entail limitations such as low-precision and limited range
of the parameters and restrictions on the size and topology of the RBM. We
conduct software simulations to determine how harmful each of these
restrictions is. Our simulations are designed to reproduce aspects of the
D-Wave quantum computer, but the issues we investigate arise in most forms of
physical computation.
|
Classification of Human Ventricular Arrhythmia in High Dimensional
Representation Spaces | We studied classification of human ECGs labelled as normal sinus rhythm,
ventricular fibrillation and ventricular tachycardia by means of support vector
machines in different representation spaces, using different observation
lengths. ECG waveform segments of duration 0.5-4 s, their Fourier magnitude
spectra, and lower dimensional projections of Fourier magnitude spectra were
used for classification. All considered representations were of much higher
dimension than in published studies. Classification accuracy improved with
segment duration up to 2 s, with 4 s providing little improvement. We found
that it is possible to discriminate between ventricular tachycardia and
ventricular fibrillation by the present approach with much shorter runs of ECG
(2 s, minimum 86% sensitivity per class) than previously imagined. Ensembles of
classifiers acting on 1 s segments taken over 5 s observation windows gave best
results, with sensitivities of detection for all classes exceeding 93%.
|
Missing Value Imputation With Unsupervised Backpropagation | Many data mining and data analysis techniques operate on dense matrices or
complete tables of data. Real-world data sets, however, often contain unknown
values. Even many classification algorithms that are designed to operate with
missing values still exhibit deteriorated accuracy. One approach to handling
missing values is to fill in (impute) the missing values. In this paper, we
present a technique for unsupervised learning called Unsupervised
Backpropagation (UBP), which trains a multi-layer perceptron to fit to the
manifold sampled by a set of observed point-vectors. We evaluate UBP with the
task of imputing missing values in datasets, and show that UBP is able to
predict missing values with significantly lower sum-squared error than other
collaborative filtering and imputation techniques. We also demonstrate with 24
datasets and 9 supervised learning algorithms that classification accuracy is
usually higher when randomly-withheld values are imputed using UBP, rather than
with other methods.
|
Continuous Learning: Engineering Super Features With Feature Algebras | In this paper we consider a problem of searching a space of predictive models
for a given training data set. We propose an iterative procedure for deriving a
sequence of improving models and a corresponding sequence of sets of non-linear
features on the original input space. After a finite number of iterations N,
the non-linear features become 2^N -degree polynomials on the original space.
We show that in a limit of an infinite number of iterations derived non-linear
features must form an associative algebra: a product of two features is equal
to a linear combination of features from the same feature space for any given
input point. Because each iteration consists of solving a series of convex
problems that contain all previous solutions, the likelihood of the models in
the sequence is increasing with each iteration while the dimension of the model
parameter space is set to a limited controlled value.
|
Approximated Infomax Early Stopping: Revisiting Gaussian RBMs on Natural
Images | We pursue an early stopping technique that helps Gaussian Restricted
Boltzmann Machines (GRBMs) to gain good natural image representations in terms
of overcompleteness and data fitting. GRBMs are widely considered as an
unsuitable model for natural images because they gain non-overcomplete
representations which include uniform filters that do not represent useful
image features. We have recently found that GRBMs once gain and subsequently
lose useful filters during their training, contrary to this common perspective.
We attribute this phenomenon to a tradeoff between overcompleteness of GRBM
representations and data fitting. To gain GRBM representations that are
overcomplete and fit data well, we propose a measure for GRBM representation
quality, approximated mutual information, and an early stopping technique based
on this measure. The proposed method boosts performance of classifiers trained
on GRBM representations.
|
Large-scale Multi-label Text Classification - Revisiting Neural Networks | Neural networks have recently been proposed for multi-label classification
because they are able to capture and model label dependencies in the output
layer. In this work, we investigate limitations of BP-MLL, a neural network
(NN) architecture that aims at minimizing pairwise ranking error. Instead, we
propose to use a comparably simple NN approach with recently proposed learning
techniques for large-scale multi-label text classification tasks. In
particular, we show that BP-MLL's ranking loss minimization can be efficiently
and effectively replaced with the commonly used cross entropy error function,
and demonstrate that several advances in neural network training that have been
developed in the realm of deep learning can be effectively employed in this
setting. Our experimental results show that simple NN models equipped with
advanced techniques such as rectified linear units, dropout, and AdaGrad
perform as well as or even outperform state-of-the-art approaches on six
large-scale textual datasets with diverse characteristics.
|
Asynchronous Adaptation and Learning over Networks --- Part I: Modeling
and Stability Analysis | In this work and the supporting Parts II [2] and III [3], we provide a rather
detailed analysis of the stability and performance of asynchronous strategies
for solving distributed optimization and adaptation problems over networks. We
examine asynchronous networks that are subject to fairly general sources of
uncertainties, such as changing topologies, random link failures, random data
arrival times, and agents turning on and off randomly. Under this model, agents
in the network may stop updating their solutions or may stop sending or
receiving information in a random manner and without coordination with other
agents. We establish in Part I conditions on the first and second-order moments
of the relevant parameter distributions to ensure mean-square stable behavior.
We derive in Part II expressions that reveal how the various parameters of the
asynchronous behavior influence network performance. We compare in Part III the
performance of asynchronous networks to the performance of both centralized
solutions and synchronous networks. One notable conclusion is that the
mean-square-error performance of asynchronous networks shows a degradation only
of the order of $O(\nu)$, where $\nu$ is a small step-size parameter, while the
convergence rate remains largely unaltered. The results provide a solid
justification for the remarkable resilience of cooperative networks in the face
of random failures at multiple levels: agents, links, data arrivals, and
topology.
|
Asynchronous Adaptation and Learning over Networks - Part II:
Performance Analysis | In Part I \cite{Zhao13TSPasync1}, we introduced a fairly general model for
asynchronous events over adaptive networks including random topologies, random
link failures, random data arrival times, and agents turning on and off
randomly. We performed a stability analysis and established the notable fact
that the network is still able to converge in the mean-square-error sense to
the desired solution. Once stable behavior is guaranteed, it becomes important
to evaluate how fast the iterates converge and how close they get to the
optimal solution. This is a demanding task due to the various asynchronous
events and due to the fact that agents influence each other. In this Part II,
we carry out a detailed analysis of the mean-square-error performance of
asynchronous strategies for solving distributed optimization and adaptation
problems over networks. We derive analytical expressions for the mean-square
convergence rate and the steady-state mean-square-deviation. The expressions
reveal how the various parameters of the asynchronous behavior influence
network performance. In the process, we establish the interesting conclusion
that even under the influence of asynchronous events, all agents in the
adaptive network can still reach an $O(\nu^{1 + \gamma_o'})$ near-agreement
with some $\gamma_o' > 0$ while approaching the desired solution within
$O(\nu)$ accuracy, where $\nu$ is proportional to the small step-size parameter
for adaptation.
|
Asynchronous Adaptation and Learning over Networks - Part III:
Comparison Analysis | In Part II [3] we carried out a detailed mean-square-error analysis of the
performance of asynchronous adaptation and learning over networks under a
fairly general model for asynchronous events including random topologies,
random link failures, random data arrival times, and agents turning on and off
randomly. In this Part III, we compare the performance of synchronous and
asynchronous networks. We also compare the performance of decentralized
adaptation against centralized stochastic-gradient (batch) solutions. Two
interesting conclusions stand out. First, the results establish that the
performance of adaptive networks is largely immune to the effect of
asynchronous events: the mean and mean-square convergence rates and the
asymptotic bias values are not degraded relative to synchronous or centralized
implementations. Only the steady-state mean-square-deviation suffers a
degradation in the order of $\nu$, which represents the small step-size
parameters used for adaptation. Second, the results show that the adaptive
distributed network matches the performance of the centralized solution. These
conclusions highlight another critical benefit of cooperation by networked
agents: cooperation does not only enhance performance in comparison to
stand-alone single-agent processing, but it also endows the network with
remarkable resilience to various forms of random failure events and is able to
deliver performance that is as powerful as batch solutions.
|
Codebook based Audio Feature Representation for Music Information
Retrieval | Digital music has become prolific in the web in recent decades. Automated
recommendation systems are essential for users to discover music they love and
for artists to reach appropriate audience. When manual annotations and user
preference data is lacking (e.g. for new artists) these systems must rely on
\emph{content based} methods. Besides powerful machine learning tools for
classification and retrieval, a key component for successful recommendation is
the \emph{audio content representation}.
Good representations should capture informative musical patterns in the audio
signal of songs. These representations should be concise, to enable efficient
(low storage, easy indexing, fast search) management of huge music
repositories, and should also be easy and fast to compute, to enable real-time
interaction with a user supplying new songs to the system.
Before designing new audio features, we explore the usage of traditional
local features, while adding a stage of encoding with a pre-computed
\emph{codebook} and a stage of pooling to get compact vectorial
representations. We experiment with different encoding methods, namely
\emph{the LASSO}, \emph{vector quantization (VQ)} and \emph{cosine similarity
(CS)}. We evaluate the representations' quality in two music information
retrieval applications: query-by-tag and query-by-example. Our results show
that concise representations can be used for successful performance in both
applications. We recommend using top-$\tau$ VQ encoding, which consistently
performs well in both applications, and requires much less computation time
than the LASSO.
|
Learning rates of $l^q$ coefficient regularization learning with
Gaussian kernel | Regularization is a well recognized powerful strategy to improve the
performance of a learning machine and $l^q$ regularization schemes with
$0<q<\infty$ are central in use. It is known that different $q$ leads to
different properties of the deduced estimators, say, $l^2$ regularization leads
to smooth estimators while $l^1$ regularization leads to sparse estimators.
Then, how does the generalization capabilities of $l^q$ regularization learning
vary with $q$? In this paper, we study this problem in the framework of
statistical learning theory and show that implementing $l^q$ coefficient
regularization schemes in the sample dependent hypothesis space associated with
Gaussian kernel can attain the same almost optimal learning rates for all
$0<q<\infty$. That is, the upper and lower bounds of learning rates for $l^q$
regularization learning are asymptotically identical for all $0<q<\infty$. Our
finding tentatively reveals that, in some modeling contexts, the choice of $q$
might not have a strong impact with respect to the generalization capability.
From this perspective, $q$ can be arbitrarily specified, or specified merely by
other no generalization criteria like smoothness, computational complexity,
sparsity, etc..
|
Word Emdeddings through Hellinger PCA | Word embeddings resulting from neural language models have been shown to be
successful for a large variety of NLP tasks. However, such architecture might
be difficult to train and time-consuming. Instead, we propose to drastically
simplify the word embeddings computation through a Hellinger PCA of the word
co-occurence matrix. We compare those new word embeddings with some well-known
embeddings on NER and movie review tasks and show that we can reach similar or
even better performance. Although deep learning is not really necessary for
generating good word embeddings, we show that it can provide an easy way to
adapt embeddings to specific tasks.
|
Multimodal Transitions for Generative Stochastic Networks | Generative Stochastic Networks (GSNs) have been recently introduced as an
alternative to traditional probabilistic modeling: instead of parametrizing the
data distribution directly, one parametrizes a transition operator for a Markov
chain whose stationary distribution is an estimator of the data generating
distribution. The result of training is therefore a machine that generates
samples through this Markov chain. However, the previously introduced GSN
consistency theorems suggest that in order to capture a wide class of
distributions, the transition operator in general should be multimodal,
something that has not been done before this paper. We introduce for the first
time multimodal transition distributions for GSNs, in particular using models
in the NADE family (Neural Autoregressive Density Estimator) as output
distributions of the transition operator. A NADE model is related to an RBM
(and can thus model multimodal distributions) but its likelihood (and
likelihood gradient) can be computed easily. The parameters of the NADE are
obtained as a learned function of the previous state of the learned Markov
chain. Experiments clearly illustrate the advantage of such multimodal
transition distributions over unimodal GSNs.
|
Playing Atari with Deep Reinforcement Learning | We present the first deep learning model to successfully learn control
policies directly from high-dimensional sensory input using reinforcement
learning. The model is a convolutional neural network, trained with a variant
of Q-learning, whose input is raw pixels and whose output is a value function
estimating future rewards. We apply our method to seven Atari 2600 games from
the Arcade Learning Environment, with no adjustment of the architecture or
learning algorithm. We find that it outperforms all previous approaches on six
of the games and surpasses a human expert on three of them.
|
Learning Transformations for Classification Forests | This work introduces a transformation-based learner model for classification
forests. The weak learner at each split node plays a crucial role in a
classification tree. We propose to optimize the splitting objective by learning
a linear transformation on subspaces using nuclear norm as the optimization
criteria. The learned linear transformation restores a low-rank structure for
data from the same class, and, at the same time, maximizes the separation
between different classes, thereby improving the performance of the split
function. Theoretical and experimental results support the proposed framework.
|
Zero-Shot Learning by Convex Combination of Semantic Embeddings | Several recent publications have proposed methods for mapping images into
continuous semantic embedding spaces. In some cases the embedding space is
trained jointly with the image transformation. In other cases the semantic
embedding space is established by an independent natural language processing
task, and then the image transformation into that space is learned in a second
stage. Proponents of these image embedding systems have stressed their
advantages over the traditional \nway{} classification framing of image
understanding, particularly in terms of the promise for zero-shot learning --
the ability to correctly annotate images of previously unseen object
categories. In this paper, we propose a simple method for constructing an image
embedding system from any existing \nway{} image classifier and a semantic word
embedding model, which contains the $\n$ class labels in its vocabulary. Our
method maps images into the semantic embedding space via convex combination of
the class label embedding vectors, and requires no additional training. We show
that this simple and direct method confers many of the advantages associated
with more complex image embedding schemes, and indeed outperforms state of the
art methods on the ImageNet zero-shot learning task.
|
k-Sparse Autoencoders | Recently, it has been observed that when representations are learnt in a way
that encourages sparsity, improved performance is obtained on classification
tasks. These methods involve combinations of activation functions, sampling
steps and different kinds of penalties. To investigate the effectiveness of
sparsity by itself, we propose the k-sparse autoencoder, which is an
autoencoder with linear activation function, where in hidden layers only the k
highest activities are kept. When applied to the MNIST and NORB datasets, we
find that this method achieves better classification results than denoising
autoencoders, networks trained with dropout, and RBMs. k-sparse autoencoders
are simple to train and the encoding stage is very fast, making them
well-suited to large problem sizes, where conventional sparse coding algorithms
cannot be applied.
|
Using Web Co-occurrence Statistics for Improving Image Categorization | Object recognition and localization are important tasks in computer vision.
The focus of this work is the incorporation of contextual information in order
to improve object recognition and localization. For instance, it is natural to
expect not to see an elephant to appear in the middle of an ocean. We consider
a simple approach to encapsulate such common sense knowledge using
co-occurrence statistics from web documents. By merely counting the number of
times nouns (such as elephants, sharks, oceans, etc.) co-occur in web
documents, we obtain a good estimate of expected co-occurrences in visual data.
We then cast the problem of combining textual co-occurrence statistics with the
predictions of image-based classifiers as an optimization problem. The
resulting optimization problem serves as a surrogate for our inference
procedure. Albeit the simplicity of the resulting optimization problem, it is
effective in improving both recognition and localization accuracy. Concretely,
we observe significant improvements in recognition and localization rates for
both ImageNet Detection 2012 and Sun 2012 datasets.
|
Time-varying Learning and Content Analytics via Sparse Factor Analysis | We propose SPARFA-Trace, a new machine learning-based framework for
time-varying learning and content analytics for education applications. We
develop a novel message passing-based, blind, approximate Kalman filter for
sparse factor analysis (SPARFA), that jointly (i) traces learner concept
knowledge over time, (ii) analyzes learner concept knowledge state transitions
(induced by interacting with learning resources, such as textbook sections,
lecture videos, etc, or the forgetting effect), and (iii) estimates the content
organization and intrinsic difficulty of the assessment questions. These
quantities are estimated solely from binary-valued (correct/incorrect) graded
learner response data and a summary of the specific actions each learner
performs (e.g., answering a question or studying a learning resource) at each
time instance. Experimental results on two online course datasets demonstrate
that SPARFA-Trace is capable of tracing each learner's concept knowledge
evolution over time, as well as analyzing the quality and content organization
of learning resources, the question-concept associations, and the question
intrinsic difficulties. Moreover, we show that SPARFA-Trace achieves comparable
or better performance in predicting unobserved learner responses than existing
collaborative filtering and knowledge tracing approaches for personalized
education.
|
SOMz: photometric redshift PDFs with self organizing maps and random
atlas | In this paper we explore the applicability of the unsupervised machine
learning technique of Self Organizing Maps (SOM) to estimate galaxy photometric
redshift probability density functions (PDFs). This technique takes a
spectroscopic training set, and maps the photometric attributes, but not the
redshifts, to a two dimensional surface by using a process of competitive
learning where neurons compete to more closely resemble the training data
multidimensional space. The key feature of a SOM is that it retains the
topology of the input set, revealing correlations between the attributes that
are not easily identified. We test three different 2D topological mapping:
rectangular, hexagonal, and spherical, by using data from the DEEP2 survey. We
also explore different implementations and boundary conditions on the map and
also introduce the idea of a random atlas where a large number of different
maps are created and their individual predictions are aggregated to produce a
more robust photometric redshift PDF. We also introduced a new metric, the
$I$-score, which efficiently incorporates different metrics, making it easier
to compare different results (from different parameters or different
photometric redshift codes). We find that by using a spherical topology mapping
we obtain a better representation of the underlying multidimensional topology,
which provides more accurate results that are comparable to other,
state-of-the-art machine learning algorithms. Our results illustrate that
unsupervised approaches have great potential for many astronomical problems,
and in particular for the computation of photometric redshifts.
|
Structure-Aware Dynamic Scheduler for Parallel Machine Learning | Training large machine learning (ML) models with many variables or parameters
can take a long time if one employs sequential procedures even with stochastic
updates. A natural solution is to turn to distributed computing on a cluster;
however, naive, unstructured parallelization of ML algorithms does not usually
lead to a proportional speedup and can even result in divergence, because
dependencies between model elements can attenuate the computational gains from
parallelization and compromise correctness of inference. Recent efforts toward
this issue have benefited from exploiting the static, a priori block structures
residing in ML algorithms. In this paper, we take this path further by
exploring the dynamic block structures and workloads therein present during ML
program execution, which offers new opportunities for improving convergence,
correctness, and load balancing in distributed ML. We propose and showcase a
general-purpose scheduler, STRADS, for coordinating distributed updates in ML
algorithms, which harnesses the aforementioned opportunities in a systematic
way. We provide theoretical guarantees for our scheduler, and demonstrate its
efficacy versus static block structures on Lasso and Matrix Factorization.
|
Consistency of Causal Inference under the Additive Noise Model | We analyze a family of methods for statistical causal inference from sample
under the so-called Additive Noise Model. While most work on the subject has
concentrated on establishing the soundness of the Additive Noise Model, the
statistical consistency of the resulting inference methods has received little
attention. We derive general conditions under which the given family of
inference methods consistently infers the causal direction in a nonparametric
setting.
|
Unsupervised Feature Learning by Deep Sparse Coding | In this paper, we propose a new unsupervised feature learning framework,
namely Deep Sparse Coding (DeepSC), that extends sparse coding to a multi-layer
architecture for visual object recognition tasks. The main innovation of the
framework is that it connects the sparse-encoders from different layers by a
sparse-to-dense module. The sparse-to-dense module is a composition of a local
spatial pooling step and a low-dimensional embedding process, which takes
advantage of the spatial smoothness information in the image. As a result, the
new method is able to learn several levels of sparse representation of the
image which capture features at a variety of abstraction levels and
simultaneously preserve the spatial smoothness between the neighboring image
patches. Combining the feature representations from multiple layers, DeepSC
achieves the state-of-the-art performance on multiple object recognition tasks.
|
Unsupervised Pretraining Encourages Moderate-Sparseness | It is well known that direct training of deep neural networks will generally
lead to poor results. A major progress in recent years is the invention of
various pretraining methods to initialize network parameters and it was shown
that such methods lead to good prediction performance. However, the reason for
the success of pretraining has not been fully understood, although it was
argued that regularization and better optimization play certain roles. This
paper provides another explanation for the effectiveness of pretraining, where
we show pretraining leads to a sparseness of hidden unit activation in the
resulting neural networks. The main reason is that the pretraining models can
be interpreted as an adaptive sparse coding. Compared to deep neural network
with sigmoid function, our experimental results on MNIST and Birdsong further
support this sparseness observation.
|
Competitive Learning with Feedforward Supervisory Signal for Pre-trained
Multilayered Networks | We propose a novel learning method for multilayered neural networks which
uses feedforward supervisory signal and associates classification of a new
input with that of pre-trained input. The proposed method effectively uses rich
input information in the earlier layer for robust leaning and revising internal
representation in a multilayer neural network.
|
Deep learning for neuroimaging: a validation study | Deep learning methods have recently made notable advances in the tasks of
classification and representation learning. These tasks are important for brain
imaging and neuroscience discovery, making the methods attractive for porting
to a neuroimager's toolbox. Success of these methods is, in part, explained by
the flexibility of deep learning models. However, this flexibility makes the
process of porting to new areas a difficult parameter optimization problem. In
this work we demonstrate our results (and feasible parameter ranges) in
application of deep learning methods to structural and functional brain imaging
data. We also describe a novel constraint-based approach to visualizing high
dimensional data. We use it to analyze the effect of parameter choices on data
transformations. Our results show that deep learning methods are able to learn
physiologically important representations and detect latent relations in
neuroimaging data.
|
Fast Training of Convolutional Networks through FFTs | Convolutional networks are one of the most widely employed architectures in
computer vision and machine learning. In order to leverage their ability to
learn complex functions, large amounts of data are required for training.
Training a large convolutional network to produce state-of-the-art results can
take weeks, even when using modern GPUs. Producing labels using a trained
network can also be costly when dealing with web-scale datasets. In this work,
we present a simple algorithm which accelerates training and inference by a
significant factor, and can yield improvements of over an order of magnitude
compared to existing state-of-the-art implementations. This is done by
computing convolutions as pointwise products in the Fourier domain while
reusing the same transformed feature map many times. The algorithm is
implemented on a GPU architecture and addresses a number of related challenges.
|
Multi-GPU Training of ConvNets | In this work we evaluate different approaches to parallelize computation of
convolutional neural networks across several GPUs.
|
A Generative Product-of-Filters Model of Audio | We propose the product-of-filters (PoF) model, a generative model that
decomposes audio spectra as sparse linear combinations of "filters" in the
log-spectral domain. PoF makes similar assumptions to those used in the classic
homomorphic filtering approach to signal processing, but replaces hand-designed
decompositions built of basic signal processing operations with a learned
decomposition based on statistical inference. This paper formulates the PoF
model and derives a mean-field method for posterior inference and a variational
EM algorithm to estimate the model's free parameters. We demonstrate PoF's
potential for audio processing on a bandwidth expansion task, and show that PoF
can serve as an effective unsupervised feature extractor for a speaker
identification task.
|
Principled Non-Linear Feature Selection | Recent non-linear feature selection approaches employing greedy optimisation
of Centred Kernel Target Alignment(KTA) exhibit strong results in terms of
generalisation accuracy and sparsity. However, they are computationally
prohibitive for large datasets. We propose randSel, a randomised feature
selection algorithm, with attractive scaling properties. Our theoretical
analysis of randSel provides strong probabilistic guarantees for correct
identification of relevant features. RandSel's characteristics make it an ideal
candidate for identifying informative learned representations. We've conducted
experimentation to establish the performance of this approach, and present
encouraging results, including a 3rd position result in the recent ICML black
box learning challenge as well as competitive results for signal peptide
prediction, an important problem in bioinformatics.
|
Group-sparse Embeddings in Collective Matrix Factorization | CMF is a technique for simultaneously learning low-rank representations based
on a collection of matrices with shared entities. A typical example is the
joint modeling of user-item, item-property, and user-feature matrices in a
recommender system. The key idea in CMF is that the embeddings are shared
across the matrices, which enables transferring information between them. The
existing solutions, however, break down when the individual matrices have
low-rank structure not shared with others. In this work we present a novel CMF
solution that allows each of the matrices to have a separate low-rank structure
that is independent of the other matrices, as well as structures that are
shared only by a subset of them. We compare MAP and variational Bayesian
solutions based on alternating optimization algorithms and show that the model
automatically infers the nature of each factor using group-wise sparsity. Our
approach supports in a principled way continuous, binary and count observations
and is efficient for sparse matrices involving missing data. We illustrate the
solution on a number of examples, focusing in particular on an interesting
use-case of augmented multi-view learning.
|
Adaptive Seeding for Gaussian Mixture Models | We present new initialization methods for the expectation-maximization
algorithm for multivariate Gaussian mixture models. Our methods are adaptions
of the well-known $K$-means++ initialization and the Gonzalez algorithm.
Thereby we aim to close the gap between simple random, e.g. uniform, and
complex methods, that crucially depend on the right choice of hyperparameters.
Our extensive experiments indicate the usefulness of our methods compared to
common techniques and methods, which e.g. apply the original $K$-means++ and
Gonzalez directly, with respect to artificial as well as real-world data sets.
|
Learning Type-Driven Tensor-Based Meaning Representations | This paper investigates the learning of 3rd-order tensors representing the
semantics of transitive verbs. The meaning representations are part of a
type-driven tensor-based semantic framework, from the newly emerging field of
compositional distributional semantics. Standard techniques from the neural
networks literature are used to learn the tensors, which are tested on a
selectional preference-style task with a simple 2-dimensional sentence space.
Promising results are obtained against a competitive corpus-based baseline. We
argue that extending this work beyond transitive verbs, and to
higher-dimensional sentence spaces, is an interesting and challenging problem
for the machine learning community to consider.
|
Stochastic Gradient Estimate Variance in Contrastive Divergence and
Persistent Contrastive Divergence | Contrastive Divergence (CD) and Persistent Contrastive Divergence (PCD) are
popular methods for training the weights of Restricted Boltzmann Machines.
However, both methods use an approximate method for sampling from the model
distribution. As a side effect, these approximations yield significantly
different biases and variances for stochastic gradient estimates of individual
data points. It is well known that CD yields a biased gradient estimate. In
this paper we however show empirically that CD has a lower stochastic gradient
estimate variance than exact sampling, while the mean of subsequent PCD
estimates has a higher variance than exact sampling. The results give one
explanation to the finding that CD can be used with smaller minibatches or
higher learning rates than PCD.
|
How to Construct Deep Recurrent Neural Networks | In this paper, we explore different ways to extend a recurrent neural network
(RNN) to a \textit{deep} RNN. We start by arguing that the concept of depth in
an RNN is not as clear as it is in feedforward neural networks. By carefully
analyzing and understanding the architecture of an RNN, however, we find three
points of an RNN which may be made deeper; (1) input-to-hidden function, (2)
hidden-to-hidden transition and (3) hidden-to-output function. Based on this
observation, we propose two novel architectures of a deep RNN which are
orthogonal to an earlier attempt of stacking multiple recurrent layers to build
a deep RNN (Schmidhuber, 1992; El Hihi and Bengio, 1996). We provide an
alternative interpretation of these deep RNNs using a novel framework based on
neural operators. The proposed deep RNNs are empirically evaluated on the tasks
of polyphonic music prediction and language modeling. The experimental result
supports our claim that the proposed deep RNNs benefit from the depth and
outperform the conventional, shallow RNNs.
|
Learning States Representations in POMDP | We propose to deal with sequential processes where only partial observations
are available by learning a latent representation space on which policies may
be accurately learned.
|
Unit Tests for Stochastic Optimization | Optimization by stochastic gradient descent is an important component of many
large-scale machine learning algorithms. A wide variety of such optimization
algorithms have been devised; however, it is unclear whether these algorithms
are robust and widely applicable across many different optimization landscapes.
In this paper we develop a collection of unit tests for stochastic
optimization. Each unit test rapidly evaluates an optimization algorithm on a
small-scale, isolated, and well-understood difficulty, rather than in
real-world scenarios where many such issues are entangled. Passing these unit
tests is not sufficient, but absolutely necessary for any algorithms with
claims to generality or robustness. We give initial quantitative and
qualitative results on numerous established algorithms. The testing framework
is open-source, extensible, and easy to apply to new algorithms.
|
Stopping Criteria in Contrastive Divergence: Alternatives to the
Reconstruction Error | Restricted Boltzmann Machines (RBMs) are general unsupervised learning
devices to ascertain generative models of data distributions. RBMs are often
trained using the Contrastive Divergence learning algorithm (CD), an
approximation to the gradient of the data log-likelihood. A simple
reconstruction error is often used to decide whether the approximation provided
by the CD algorithm is good enough, though several authors (Schulz et al.,
2010; Fischer & Igel, 2010) have raised doubts concerning the feasibility of
this procedure. However, not many alternatives to the reconstruction error have
been used in the literature. In this manuscript we investigate simple
alternatives to the reconstruction error in order to detect as soon as possible
the decrease in the log-likelihood during learning.
|
The return of AdaBoost.MH: multi-class Hamming trees | Within the framework of AdaBoost.MH, we propose to train vector-valued
decision trees to optimize the multi-class edge without reducing the
multi-class problem to $K$ binary one-against-all classifications. The key
element of the method is a vector-valued decision stump, factorized into an
input-independent vector of length $K$ and label-independent scalar classifier.
At inner tree nodes, the label-dependent vector is discarded and the binary
classifier can be used for partitioning the input space into two regions. The
algorithm retains the conceptual elegance, power, and computational efficiency
of binary AdaBoost. In experiments it is on par with support vector machines
and with the best existing multi-class boosting algorithm AOSOLogitBoost, and
it is significantly better than other known implementations of AdaBoost.MH.
|
On the number of response regions of deep feed forward networks with
piece-wise linear activations | This paper explores the complexity of deep feedforward networks with linear
pre-synaptic couplings and rectified linear activations. This is a contribution
to the growing body of work contrasting the representational power of deep and
shallow network architectures. In particular, we offer a framework for
comparing deep and shallow models that belong to the family of piecewise linear
functions based on computational geometry. We look at a deep rectifier
multi-layer perceptron (MLP) with linear outputs units and compare it with a
single layer version of the model. In the asymptotic regime, when the number of
inputs stays constant, if the shallow model has $kn$ hidden units and $n_0$
inputs, then the number of linear regions is $O(k^{n_0}n^{n_0})$. For a $k$
layer model with $n$ hidden units on each layer it is $\Omega(\left\lfloor
{n}/{n_0}\right\rfloor^{k-1}n^{n_0})$. The number
$\left\lfloor{n}/{n_0}\right\rfloor^{k-1}$ grows faster than $k^{n_0}$ when $n$
tends to infinity or when $k$ tends to infinity and $n \geq 2n_0$.
Additionally, even when $k$ is small, if we restrict $n$ to be $2n_0$, we can
show that a deep model has considerably more linear regions that a shallow one.
We consider this as a first step towards understanding the complexity of these
models and specifically towards providing suitable mathematical tools for
future analysis.
|
Modeling correlations in spontaneous activity of visual cortex with
centered Gaussian-binary deep Boltzmann machines | Spontaneous cortical activity -- the ongoing cortical activities in absence
of intentional sensory input -- is considered to play a vital role in many
aspects of both normal brain functions and mental dysfunctions. We present a
centered Gaussian-binary Deep Boltzmann Machine (GDBM) for modeling the
activity in early cortical visual areas and relate the random sampling in GDBMs
to the spontaneous cortical activity. After training the proposed model on
natural image patches, we show that the samples collected from the model's
probability distribution encompass similar activity patterns as found in the
spontaneous activity. Specifically, filters having the same orientation
preference tend to be active together during random sampling. Our work
demonstrates the centered GDBM is a meaningful model approach for basic
receptive field properties and the emergence of spontaneous activity patterns
in early cortical visual areas. Besides, we show empirically that centered
GDBMs do not suffer from the difficulties during training as GDBMs do and can
be properly trained without the layer-wise pretraining.
|
Auto-Encoding Variational Bayes | How can we perform efficient inference and learning in directed probabilistic
models, in the presence of continuous latent variables with intractable
posterior distributions, and large datasets? We introduce a stochastic
variational inference and learning algorithm that scales to large datasets and,
under some mild differentiability conditions, even works in the intractable
case. Our contributions is two-fold. First, we show that a reparameterization
of the variational lower bound yields a lower bound estimator that can be
straightforwardly optimized using standard stochastic gradient methods. Second,
we show that for i.i.d. datasets with continuous latent variables per
datapoint, posterior inference can be made especially efficient by fitting an
approximate inference model (also called a recognition model) to the
intractable posterior using the proposed lower bound estimator. Theoretical
advantages are reflected in experimental results.
|
Neuronal Synchrony in Complex-Valued Deep Networks | Deep learning has recently led to great successes in tasks such as image
recognition (e.g Krizhevsky et al., 2012). However, deep networks are still
outmatched by the power and versatility of the brain, perhaps in part due to
the richer neuronal computations available to cortical circuits. The challenge
is to identify which neuronal mechanisms are relevant, and to find suitable
abstractions to model them. Here, we show how aspects of spike timing, long
hypothesized to play a crucial role in cortical information processing, could
be incorporated into deep networks to build richer, versatile representations.
We introduce a neural network formulation based on complex-valued neuronal
units that is not only biologically meaningful but also amenable to a variety
of deep learning frameworks. Here, units are attributed both a firing rate and
a phase, the latter indicating properties of spike timing. We show how this
formulation qualitatively captures several aspects thought to be related to
neuronal synchrony, including gating of information processing and dynamic
binding of distributed object representations. Focusing on the latter, we
demonstrate the potential of the approach in several simple experiments. Thus,
neuronal synchrony could be a flexible mechanism that fulfills multiple
functional roles in deep networks.
|
Subsets and Splits
No saved queries yet
Save your SQL queries to embed, download, and access them later. Queries will appear here once saved.