title
stringlengths 7
246
| abstract
stringlengths 6
3.31k
|
|---|---|
Optional Stopping with Bayes Factors: a categorization and extension of
folklore results, with an application to invariant situations
|
It is often claimed that Bayesian methods, in particular Bayes factor methods
for hypothesis testing, can deal with optional stopping. We first give an
overview, using elementary probability theory, of three different mathematical
meanings that various authors give to this claim: (1) stopping rule
independence, (2) posterior calibration and (3) (semi-) frequentist robustness
to optional stopping. We then prove theorems to the effect that these claims do
indeed hold in a general measure-theoretic setting. For claims of type (2) and
(3), such results are new. By allowing for non-integrable measures based on
improper priors, we obtain particularly strong results for the practically
important case of models with nuisance parameters satisfying a group invariance
(such as location or scale). We also discuss the practical relevance of
(1)--(3), and conclude that whether Bayes factor methods actually perform well
under optional stopping crucially depends on details of models, priors and the
goal of the analysis.
|
Decision Variance in Online Learning
|
Online learning has traditionally focused on the expected rewards. In this
paper, a risk-averse online learning problem under the performance measure of
the mean-variance of the rewards is studied. Both the bandit and full
information settings are considered. The performance of several existing
policies is analyzed, and new fundamental limitations on risk-averse learning
is established. In particular, it is shown that although a logarithmic
distribution-dependent regret in time $T$ is achievable (similar to the
risk-neutral problem), the worst-case (i.e. minimax) regret is lower bounded by
$\Omega(T)$ (in contrast to the $\Omega(\sqrt{T})$ lower bound in the
risk-neutral problem). This sharp difference from the risk-neutral counterpart
is caused by the the variance in the player's decisions, which, while absent in
the regret under the expected reward criterion, contributes to excess
mean-variance due to the non-linearity of this risk measure. The role of the
decision variance in regret performance reflects a risk-averse player's desire
for robust decisions and outcomes.
|
Algorithm Selection for Collaborative Filtering: the influence of graph
metafeatures and multicriteria metatargets
|
To select the best algorithm for a new problem is an expensive and difficult
task. However, there are automatic solutions to address this problem: using
Metalearning, which takes advantage of problem characteristics (i.e.
metafeatures), one is able to predict the relative performance of algorithms.
In the Collaborative Filtering scope, recent works have proposed diverse
metafeatures describing several dimensions of this problem. Despite interesting
and effective findings, it is still unknown whether these are the most
effective metafeatures. Hence, this work proposes a new set of graph
metafeatures, which approach the Collaborative Filtering problem from a Graph
Theory perspective. Furthermore, in order to understand whether metafeatures
from multiple dimensions are a better fit, we investigate the effects of
comprehensive metafeatures. These metafeatures are a selection of the best
metafeatures from all existing Collaborative Filtering metafeatures. The impact
of the most representative metafeatures is investigated in a controlled
experimental setup. Another contribution we present is the use of a
Pareto-Efficient ranking procedure to create multicriteria metatargets. These
new rankings of algorithms, which take into account multiple evaluation
measures, allow to explore the algorithm selection problem in a fairer and more
detailed way. According to the experimental results, the graph metafeatures are
a good alternative to related work metafeatures. However, the results have
shown that the feature selection procedure used to create the comprehensive
metafeatures is is not effective, since there is no gain in predictive
performance. Finally, an extensive metaknowledge analysis was conducted to
identify the most influential metafeatures.
|
A Structured Learning Approach with Neural Conditional Random Fields for
Sleep Staging
|
Sleep plays a vital role in human health, both mental and physical. Sleep
disorders like sleep apnea are increasing in prevalence, with the rapid
increase in factors like obesity. Sleep apnea is most commonly treated with
Continuous Positive Air Pressure (CPAP) therapy. Presently, however, there is
no mechanism to monitor a patient's progress with CPAP. Accurate detection of
sleep stages from CPAP flow signal is crucial for such a mechanism. We propose,
for the first time, an automated sleep staging model based only on the flow
signal. Deep neural networks have recently shown high accuracy on sleep staging
by eliminating handcrafted features. However, these methods focus exclusively
on extracting informative features from the input signal, without paying much
attention to the dynamics of sleep stages in the output sequence. We propose an
end-to-end framework that uses a combination of deep convolution and recurrent
neural networks to extract high-level features from raw flow signal with a
structured output layer based on a conditional random field to model the
temporal transition structure of the sleep stages. We improve upon the previous
methods by 10% using our model, that can be augmented to the previous sleep
staging deep learning methods. We also show that our method can be used to
accurately track sleep metrics like sleep efficiency calculated from sleep
stages that can be deployed for monitoring the response of CPAP therapy on
sleep apnea patients. Apart from the technical contributions, we expect this
study to motivate new research questions in sleep science.
|
Recurrent Neural Networks for Long and Short-Term Sequential
Recommendation
|
Recommender systems objectives can be broadly characterized as modeling user
preferences over short-or long-term time horizon. A large body of previous
research studied long-term recommendation through dimensionality reduction
techniques applied to the historical user-item interactions. A recently
introduced session-based recommendation setting highlighted the importance of
modeling short-term user preferences. In this task, Recurrent Neural Networks
(RNN) have shown to be successful at capturing the nuances of user's
interactions within a short time window. In this paper, we evaluate RNN-based
models on both short-term and long-term recommendation tasks. Our experimental
results suggest that RNNs are capable of predicting immediate as well as
distant user interactions. We also find the best performing configuration to be
a stacked RNN with layer normalization and tied item embeddings.
|
Clearing noisy annotations for computed tomography imaging
|
One of the problems on the way to successful implementation of neural
networks is the quality of annotation. For instance, different annotators can
annotate images in a different way and very often their decisions do not match
exactly and in extreme cases are even mutually exclusive which results in noisy
annotations and, consequently, inaccurate predictions.
To avoid that problem in the task of computed tomography (CT) imaging
segmentation we propose a clearing algorithm for annotations. It consists of 3
stages:
- annotators scoring, which assigns a higher confidence level to better
annotators;
- nodules scoring, which assigns a higher confidence level to nodules
confirmed by good annotators;
- nodules merging, which aggregates annotations according to nodules
confidence.
In general, the algorithm can be applied to many different tasks (namely,
binary and multi-class semantic segmentation, and also with trivial adjustments
to classification and regression) where there are several annotators labeling
each image.
|
An argument in favor of strong scaling for deep neural networks with
small datasets
|
In recent years, with the popularization of deep learning frameworks and
large datasets, researchers have started parallelizing their models in order to
train faster. This is crucially important, because they typically explore many
hyperparameters in order to find the best ones for their applications. This
process is time consuming and, consequently, speeding up training improves
productivity. One approach to parallelize deep learning models followed by many
researchers is based on weak scaling. The minibatches increase in size as new
GPUs are added to the system. In addition, new learning rates schedules have
been proposed to fix optimization issues that occur with large minibatch sizes.
In this paper, however, we show that the recommendations provided by recent
work do not apply to models that lack large datasets. In fact, we argument in
favor of using strong scaling for achieving reliable performance in such cases.
We evaluated our approach with up to 32 GPUs and show that weak scaling not
only does not have the same accuracy as the sequential model, it also fails to
converge most of time. Meanwhile, strong scaling has good scalability while
having exactly the same accuracy of a sequential implementation.
|
Robot Imitation through Vision, Kinesthetic and Force Features with
Online Adaptation to Changing Environments
|
Continuous Goal-Directed Actions (CGDA) is a robot imitation framework that
encodes actions as the changes they produce on the environment. While it
presents numerous advantages with respect to other robot imitation frameworks
in terms of generalization and portability, final robot joint trajectories for
the execution of actions are not necessarily encoded within the model. This is
studied as an optimization problem, and the solution is computed through
evolutionary algorithms in simulated environments. Evolutionary algorithms
require a large number of evaluations, which had made the use of these
algorithms in real world applications very challenging. This paper presents
online evolutionary strategies, as a change of paradigm within CGDA execution.
Online evolutionary strategies shift and merge motor execution into the
planning loop. A concrete online evolutionary strategy, Online Evolved
Trajectories (OET), is presented. OET drastically reduces computational times
between motor executions, and enables working in real world dynamic
environments and/or with human collaboration. Its performance has been measured
against Full Trajectory Evolution (FTE) and Incrementally Evolved Trajectories
(IET), obtaining the best overall results. Experimental evaluations are
performed on the TEO full-sized humanoid robot with "paint" and "iron" actions
that together involve vision, kinesthetic and force features.
|
Self-Paced Learning with Adaptive Deep Visual Embeddings
|
Selecting the most appropriate data examples to present a deep neural network
(DNN) at different stages of training is an unsolved challenge. Though
practitioners typically ignore this problem, a non-trivial data scheduling
method may result in a significant improvement in both convergence and
generalization performance. In this paper, we introduce Self-Paced Learning
with Adaptive Deep Visual Embeddings (SPL-ADVisE), a novel end-to-end training
protocol that unites self-paced learning (SPL) and deep metric learning (DML).
We leverage the Magnet Loss to train an embedding convolutional neural network
(CNN) to learn a salient representation space. The student CNN classifier
dynamically selects similar instance-level training examples to form a
mini-batch, where the easiness from the cross-entropy loss and the true
diverseness of examples from the learned metric space serve as sample
importance priors. To demonstrate the effectiveness of SPL-ADVisE, we use deep
CNN architectures for the task of supervised image classification on several
coarse- and fine-grained visual recognition datasets. Results show that, across
all datasets, the proposed method converges faster and reaches a higher final
accuracy than other SPL variants, particularly on fine-grained classes.
|
Constraint-Based Visual Generation
|
In the last few years the systematic adoption of deep learning to visual
generation has produced impressive results that, amongst others, definitely
benefit from the massive exploration of convolutional architectures. In this
paper, we propose a general approach to visual generation that combines
learning capabilities with logic descriptions of the target to be generated.
The process of generation is regarded as a constrained satisfaction problem,
where the constraints describe a set of properties that characterize the
target. Interestingly, the constraints can also involve logic variables, while
all of them are converted into real-valued functions by means of the t-norm
theory. We use deep architectures to model the involved variables, and propose
a computational scheme where the learning process carries out a satisfaction of
the constraints. We propose some examples in which the theory can naturally be
used, including the modeling of GAN and auto-encoders, and report promising
results in problems with the generation of handwritten characters and face
transformations.
|
Improving pairwise comparison models using Empirical Bayes shrinkage
|
Comparison data arises in many important contexts, e.g. shopping, web clicks,
or sports competitions. Typically we are given a dataset of comparisons and
wish to train a model to make predictions about the outcome of unseen
comparisons. In many cases available datasets have relatively few comparisons
(e.g. there are only so many NFL games per year) or efficiency is important
(e.g. we want to quickly estimate the relative appeal of a product). In such
settings it is well known that shrinkage estimators outperform maximum
likelihood estimators. A complicating matter is that standard comparison models
such as the conditional multinomial logit model are only models of conditional
outcomes (who wins) and not of comparisons themselves (who competes). As such,
different models of the comparison process lead to different shrinkage
estimators. In this work we derive a collection of methods for estimating the
pairwise uncertainty of pairwise predictions based on different assumptions
about the comparison process. These uncertainty estimates allow us both to
examine model uncertainty as well as perform Empirical Bayes shrinkage
estimation of the model parameters. We demonstrate that our shrunk estimators
outperform standard maximum likelihood methods on real comparison data from
online comparison surveys as well as from several sports contexts.
|
Unsupervised Learning of Latent Physical Properties Using
Perception-Prediction Networks
|
We propose a framework for the completely unsupervised learning of latent
object properties from their interactions: the perception-prediction network
(PPN). Consisting of a perception module that extracts representations of
latent object properties and a prediction module that uses those extracted
properties to simulate system dynamics, the PPN can be trained in an end-to-end
fashion purely from samples of object dynamics. The representations of latent
object properties learned by PPNs not only are sufficient to accurately
simulate the dynamics of systems comprised of previously unseen objects, but
also can be translated directly into human-interpretable properties (e.g.,
mass, coefficient of restitution) in an entirely unsupervised manner.
Crucially, PPNs also generalize to novel scenarios: their gradient-based
training can be applied to many dynamical systems and their graph-based
structure functions over systems comprised of different numbers of objects. Our
results demonstrate the efficacy of graph-based neural architectures in
object-centric inference and prediction tasks, and our model has the potential
to discover relevant object properties in systems that are not yet well
understood.
|
Visual Dynamics: Stochastic Future Generation via Layered Cross
Convolutional Networks
|
We study the problem of synthesizing a number of likely future frames from a
single input image. In contrast to traditional methods that have tackled this
problem in a deterministic or non-parametric way, we propose to model future
frames in a probabilistic manner. Our probabilistic model makes it possible for
us to sample and synthesize many possible future frames from a single input
image. To synthesize realistic movement of objects, we propose a novel network
structure, namely a Cross Convolutional Network; this network encodes image and
motion information as feature maps and convolutional kernels, respectively. In
experiments, our model performs well on synthetic data, such as 2D shapes and
animated game sprites, and on real-world video frames. We present analyses of
the learned network representations, showing it is implicitly learning a
compact encoding of object appearance and motion. We also demonstrate a few of
its applications, including visual analogy-making and video extrapolation.
|
Learning to Generate and Reconstruct 3D Meshes with only 2D Supervision
|
We present a unified framework tackling two problems: class-specific 3D
reconstruction from a single image, and generation of new 3D shape samples.
These tasks have received considerable attention recently; however, existing
approaches rely on 3D supervision, annotation of 2D images with keypoints or
poses, and/or training with multiple views of each object instance. Our
framework is very general: it can be trained in similar settings to these
existing approaches, while also supporting weaker supervision scenarios.
Importantly, it can be trained purely from 2D images, without ground-truth pose
annotations, and with a single view per instance. We employ meshes as an output
representation, instead of voxels used in most prior work. This allows us to
exploit shading information during training, which previous 2D-supervised
methods cannot. Thus, our method can learn to generate and reconstruct concave
object classes. We evaluate our approach on synthetic data in various settings,
showing that (i) it learns to disentangle shape from pose; (ii) using shading
in the loss improves performance; (iii) our model is comparable or superior to
state-of-the-art voxel-based approaches on quantitative metrics, while
producing results that are visually more pleasing; (iv) it still performs well
when given supervision weaker than in prior works.
|
Noise Contrastive Priors for Functional Uncertainty
|
Obtaining reliable uncertainty estimates of neural network predictions is a
long standing challenge. Bayesian neural networks have been proposed as a
solution, but it remains open how to specify their prior. In particular, the
common practice of an independent normal prior in weight space imposes
relatively weak constraints on the function posterior, allowing it to
generalize in unforeseen ways on inputs outside of the training distribution.
We propose noise contrastive priors (NCPs) to obtain reliable uncertainty
estimates. The key idea is to train the model to output high uncertainty for
data points outside of the training distribution. NCPs do so using an input
prior, which adds noise to the inputs of the current mini batch, and an output
prior, which is a wide distribution given these inputs. NCPs are compatible
with any model that can output uncertainty estimates, are easy to scale, and
yield reliable uncertainty estimates throughout training. Empirically, we show
that NCPs prevent overfitting outside of the training distribution and result
in uncertainty estimates that are useful for active learning. We demonstrate
the scalability of our method on the flight delays data set, where we
significantly improve upon previously published results.
|
Improved Training with Curriculum GANs
|
In this paper we introduce Curriculum GANs, a curriculum learning strategy
for training Generative Adversarial Networks that increases the strength of the
discriminator over the course of training, thereby making the learning task
progressively more difficult for the generator. We demonstrate that this
strategy is key to obtaining state-of-the-art results in image generation. We
also show evidence that this strategy may be broadly applicable to improving
GAN training in other data modalities.
|
Automatic Bayesian Density Analysis
|
Making sense of a dataset in an automatic and unsupervised fashion is a
challenging problem in statistics and AI. Classical approaches for {exploratory
data analysis} are usually not flexible enough to deal with the uncertainty
inherent to real-world data: they are often restricted to fixed latent
interaction models and homogeneous likelihoods; they are sensitive to missing,
corrupt and anomalous data; moreover, their expressiveness generally comes at
the price of intractable inference. As a result, supervision from statisticians
is usually needed to find the right model for the data. However, since domain
experts are not necessarily also experts in statistics, we propose Automatic
Bayesian Density Analysis (ABDA) to make exploratory data analysis accessible
at large. Specifically, ABDA allows for automatic and efficient missing value
estimation, statistical data type and likelihood discovery, anomaly detection
and dependency structure mining, on top of providing accurate density
estimation. Extensive empirical evidence shows that ABDA is a suitable tool for
automatic exploratory analysis of mixed continuous and discrete tabular data.
|
A Simple Probabilistic Model for Uncertainty Estimation
|
The article focuses on determining the predictive uncertainty of a model on
the example of atrial fibrillation detection problem by a single-lead ECG
signal. To this end, the model predicts parameters of the beta distribution
over class probabilities instead of these probabilities themselves. It was
shown that the described approach allows to detect atypical recordings and
significantly improve the quality of the algorithm on confident predictions.
|
Learning Plannable Representations with Causal InfoGAN
|
In recent years, deep generative models have been shown to 'imagine'
convincing high-dimensional observations such as images, audio, and even video,
learning directly from raw data. In this work, we ask how to imagine
goal-directed visual plans -- a plausible sequence of observations that
transition a dynamical system from its current configuration to a desired goal
state, which can later be used as a reference trajectory for control. We focus
on systems with high-dimensional observations, such as images, and propose an
approach that naturally combines representation learning and planning. Our
framework learns a generative model of sequential observations, where the
generative process is induced by a transition in a low-dimensional planning
model, and an additional noise. By maximizing the mutual information between
the generated observations and the transition in the planning model, we obtain
a low-dimensional representation that best explains the causal nature of the
data. We structure the planning model to be compatible with efficient planning
algorithms, and we propose several such models based on either discrete or
continuous states. Finally, to generate a visual plan, we project the current
and goal observations onto their respective states in the planning model, plan
a trajectory, and then use the generative model to transform the trajectory to
a sequence of observations. We demonstrate our method on imagining plausible
visual plans of rope manipulation.
|
Deep learning the high variability and randomness inside multimode
fibres
|
Multimode fibres (MMF) are remarkable high-capacity information channels
owing to the large number of transmitting fibre modes, and have recently
attracted significant renewed interest in applications such as optical
communication, imaging, and optical trapping. At the same time, the optical
transmitting modes inside MMFs are highly sensitive to external perturbations
and environmental changes, resulting in MMF transmission channels being highly
variable and random. This largely limits the practical application of MMFs and
hinders the full exploitation of their information capacity. Despite great
research efforts made to overcome the high variability and randomness inside
MMFs, any geometric change to the MMF leads to completely different
transmission matrices, which unavoidably fails at the information recovery.
Here, we show the successful binary image transmission using deep learning
through a single MMF, which is stationary or subject to dynamic shape
variations. We found that a single convolutional neural network has excellent
generalisation capability with various MMF transmission states. This deep
neural network can be trained by multiple MMF transmission states to accurately
predict unknown information at the other end of the MMF at any of these states,
without knowing which state is present. Our results demonstrate that deep
learning is a promising solution to address the variability and randomness
challenge of MMF based information channels. This deep-learning approach is the
starting point of developing future high-capacity MMF optical systems and
devices, and is applicable to optical systems concerning other diffusing media.
|
Iterative Amortized Inference
|
Inference models are a key component in scaling variational inference to deep
latent variable models, most notably as encoder networks in variational
auto-encoders (VAEs). By replacing conventional optimization-based inference
with a learned model, inference is amortized over data examples and therefore
more computationally efficient. However, standard inference models are
restricted to direct mappings from data to approximate posterior estimates. The
failure of these models to reach fully optimized approximate posterior
estimates results in an amortization gap. We aim toward closing this gap by
proposing iterative inference models, which learn to perform inference
optimization through repeatedly encoding gradients. Our approach generalizes
standard inference models in VAEs and provides insight into several empirical
findings, including top-down inference techniques. We demonstrate the inference
optimization capabilities of iterative inference models and show that they
outperform standard inference models on several benchmark data sets of images
and text.
|
Unsupervised Learning with Self-Organizing Spiking Neural Networks
|
We present a system comprising a hybridization of self-organized map (SOM)
properties with spiking neural networks (SNNs) that retain many of the features
of SOMs. Networks are trained in an unsupervised manner to learn a
self-organized lattice of filters via excitatory-inhibitory interactions among
populations of neurons. We develop and test various inhibition strategies, such
as growing with inter-neuron distance and two distinct levels of inhibition.
The quality of the unsupervised learning algorithm is evaluated using examples
with known labels. Several biologically-inspired classification tools are
proposed and compared, including population-level confidence rating, and
n-grams using spike motif algorithm. Using the optimal choice of parameters,
our approach produces improvements over state-of-art spiking neural networks.
|
On sampling from a log-concave density using kinetic Langevin diffusions
|
Langevin diffusion processes and their discretizations are often used for
sampling from a target density. The most convenient framework for assessing the
quality of such a sampling scheme corresponds to smooth and strongly
log-concave densities defined on $\mathbb R^p$. The present work focuses on
this framework and studies the behavior of Monte Carlo algorithms based on
discretizations of the kinetic Langevin diffusion. We first prove the geometric
mixing property of the kinetic Langevin diffusion with a mixing rate that is,
in the overdamped regime, optimal in terms of its dependence on the condition
number. We then use this result for obtaining improved guarantees of sampling
using the kinetic Langevin Monte Carlo method, when the quality of sampling is
measured by the Wasserstein distance. We also consider the situation where the
Hessian of the log-density of the target distribution is Lipschitz-continuous.
In this case, we introduce a new discretization of the kinetic Langevin
diffusion and prove that this leads to a substantial improvement of the upper
bound on the sampling error measured in Wasserstein distance.
|
Domain Stylization: A Strong, Simple Baseline for Synthetic to Real
Image Domain Adaptation
|
Deep neural networks have largely failed to effectively utilize synthetic
data when applied to real images due to the covariate shift problem. In this
paper, we show that by applying a straightforward modification to an existing
photorealistic style transfer algorithm, we achieve state-of-the-art
synthetic-to-real domain adaptation results. We conduct extensive experimental
validations on four synthetic-to-real tasks for semantic segmentation and
object detection, and show that our approach exceeds the performance of any
current state-of-the-art GAN-based image translation approach as measured by
segmentation and object detection metrics. Furthermore we offer a distance
based analysis of our method which shows a dramatic reduction in Frechet
Inception distance between the source and target domains, offering a
quantitative metric that demonstrates the effectiveness of our algorithm in
bridging the synthetic-to-real gap.
|
On the Randomized Complexity of Minimizing a Convex Quadratic Function
|
Minimizing a convex, quadratic objective of the form
$f_{\mathbf{A},\mathbf{b}}(x) := \frac{1}{2}x^\top \mathbf{A} x - \langle
\mathbf{b}, x \rangle$ for $\mathbf{A} \succ 0 $ is a fundamental problem in
machine learning and optimization. In this work, we prove gradient-query
complexity lower bounds for minimizing convex quadratic functions which apply
to both deterministic and \emph{randomized} algorithms. Specifically, for
$\kappa > 1$, we exhibit a distribution over $(\mathbf{A},\mathbf{b})$ with
condition number $\mathrm{cond}(\mathbf{A}) \le \kappa$, such that any
\emph{randomized} algorithm requires $\Omega(\sqrt{\kappa})$ gradient queries
to find a solution $\hat x$ for which $\|\hat x - \mathbf x_\star\| \le
\epsilon_0\|\mathbf{x}_{\star}\|$, where $\mathbf x_{\star} =
\mathbf{A}^{-1}\mathbf{b}$ is the optimal solution, and $\epsilon_0$ a small
constant. Setting $\kappa =1/\epsilon$, this lower bound implies the minimax
rate of $T = \Omega(\lambda_1(\mathbf{A})\|\mathbf
x_\star\|^2/\sqrt{\epsilon})$ queries required to minimize an arbitrary convex
quadratic function up to error $f(\hat{x}) - f(\mathbf x_\star) \le \epsilon$.
Our lower bound holds for a distribution derived from classical ensembles in
random matrix theory, and relies on a careful reduction from adaptively
estimating a planted vector $\mathbf u$ in a deformed Wigner model. A key step
in deriving sharp lower bounds is demonstrating that the optimization error
$\mathbf x_\star - \hat x$ cannot align too closely with $\mathbf{u}$. To this
end, we prove an upper bound on the cosine between $\mathbf x_\star - \hat x$
and $\mathbf u$ in terms of the MMSE of estimating the plant $\mathbf u$ in a
deformed Wigner model. We then bound the MMSE by carefully modifying a result
due to Lelarge and Miolane 2016, which rigorously establishes a general
replica-symmetric formula for planted matrix models.
|
Learning from Delayed Outcomes via Proxies with Applications to
Recommender Systems
|
Predicting delayed outcomes is an important problem in recommender systems
(e.g., if customers will finish reading an ebook). We formalize the problem as
an adversarial, delayed online learning problem and consider how a proxy for
the delayed outcome (e.g., if customers read a third of the book in 24 hours)
can help minimize regret, even though the proxy is not available when making a
prediction. Motivated by our regret analysis, we propose two neural network
architectures: Factored Forecaster (FF) which is ideal if the proxy is
informative of the outcome in hindsight, and Residual Factored Forecaster (RFF)
that is robust to a non-informative proxy. Experiments on two real-world
datasets for predicting human behavior show that RFF outperforms both FF and a
direct forecaster that does not make use of the proxy. Our results suggest that
exploiting proxies by factorization is a promising way to mitigate the impact
of long delays in human-behavior prediction tasks.
|
LAPRAN: A Scalable Laplacian Pyramid Reconstructive Adversarial Network
for Flexible Compressive Sensing Reconstruction
|
This paper addresses the single-image compressive sensing (CS) and
reconstruction problem. We propose a scalable Laplacian pyramid reconstructive
adversarial network (LAPRAN) that enables high-fidelity, flexible and fast CS
images reconstruction. LAPRAN progressively reconstructs an image following the
concept of Laplacian pyramid through multiple stages of reconstructive
adversarial networks (RANs). At each pyramid level, CS measurements are fused
with a contextual latent vector to generate a high-frequency image residual.
Consequently, LAPRAN can produce hierarchies of reconstructed images and each
with an incremental resolution and improved quality. The scalable pyramid
structure of LAPRAN enables high-fidelity CS reconstruction with a flexible
resolution that is adaptive to a wide range of compression ratios (CRs), which
is infeasible with existing methods. Experimental results on multiple public
datasets show that LAPRAN offers an average 7.47dB and 5.98dB PSNR, and an
average 57.93% and 33.20% SSIM improvement compared to model-based and
data-driven baselines, respectively.
|
Efficient algorithms for robust submodular maximization under matroid
constraints
|
In this work, we consider robust submodular maximization with matroid
constraints. We give an efficient bi-criteria approximation algorithm that
outputs a small family of feasible sets whose union has (nearly) optimal
objective value. This algorithm theoretically performs less function calls than
previous works at cost of adding more elements to the final solution. We also
provide significant implementation improvements showing that our algorithm
outperforms the algorithms in the existing literature. We finally assess the
performance of our contributions in three real-world applications.
|
Topics in Random Matrices and Statistical Machine Learning
|
This thesis consists of two independent parts: random matrices, which form
the first one-third of this thesis, and machine learning, which constitutes the
remaining part. The main results of this thesis are as follows: a necessary and
sufficient condition for the inverse moments of $(m,n,\beta)$-Laguerre matrices
and compound Wishart matrices to be finite; the universal weak consistency and
the strong consistency of the $k$-nearest neighbor rule in metrically
sigma-finite dimensional spaces and metrically finite dimensional spaces
respectively. In Part I, the Chapter 1 introduces the $(m,n,\beta)$-Laguerre
matrix, Wishart and compound Wishart matrix and their joint eigenvalue
distribution. While in Chapter 2, a necessary and sufficient condition to have
finite inverse moments has been derived. In Part II, the Chapter 1 introduces
the various notions of metric dimension and differentiation property followed
by our proof for the necessary part of Preiss' result. Further, Chapter 2 gives
an introduction to the mathematical concepts in statistical machine learning
and then the $k$-nearest neighbor rule is presented in Chapter 3 with a proof
of Stone's theorem. In chapters 4 and 5, we present our main results and some
possible future directions based on it.
|
Multi-Agent Reinforcement Learning: A Report on Challenges and
Approaches
|
Reinforcement Learning (RL) is a learning paradigm concerned with learning to
control a system so as to maximize an objective over the long term. This
approach to learning has received immense interest in recent times and success
manifests itself in the form of human-level performance on games like
\textit{Go}. While RL is emerging as a practical component in real-life
systems, most successes have been in Single Agent domains. This report will
instead specifically focus on challenges that are unique to Multi-Agent Systems
interacting in mixed cooperative and competitive environments. The report
concludes with advances in the paradigm of training Multi-Agent Systems called
\textit{Decentralized Actor, Centralized Critic}, based on an extension of MDPs
called \textit{Decentralized Partially Observable MDP}s, which has seen a
renewed interest lately.
|
Unbounded Output Networks for Classification
|
We proposed the expected energy-based restricted Boltzmann machine (EE-RBM)
as a discriminative RBM method for classification. Two characteristics of the
EE-RBM are that the output is unbounded and that the target value of correct
classification is set to a value much greater than one. In this study, by
adopting features of the EE-RBM approach to feed-forward neural networks, we
propose the UnBounded output network (UBnet) which is characterized by three
features: (1) unbounded output units; (2) the target value of correct
classification is set to a value much greater than one; and (3) the models are
trained by a modified mean-squared error objective. We evaluate our approach
using the MNIST, CIFAR-10, and CIFAR-100 benchmark datasets. We first
demonstrate, for shallow UBnets on MNIST, that a setting of the target value
equal to the number of hidden units significantly outperforms a setting of the
target value equal to one, and it also outperforms standard neural networks by
about 25\%. We then validate our approach by achieving high-level
classification performance on the three datasets using unbounded output
residual networks. We finally use MNIST to analyze the learned features and
weights, and we demonstrate that UBnets are much more robust against
adversarial examples than the standard approach of using a softmax output layer
and training the networks by a cross-entropy objective.
|
Propensity score estimation using classification and regression trees in
the presence of missing covariate data
|
Data mining and machine learning techniques such as classification and
regression trees (CART) represent a promising alternative to conventional
logistic regression for propensity score estimation. Whereas incomplete data
preclude the fitting of a logistic regression on all subjects, CART is
appealing in part because some implementations allow for incomplete records to
be incorporated in the tree fitting and provide propensity score estimates for
all subjects. Based on theoretical considerations, we argue that the automatic
handling of missing data by CART may however not be appropriate. Using a series
of simulation experiments, we examined the performance of different approaches
to handling missing covariate data; (i) applying the CART algorithm directly to
the (partially) incomplete data, (ii) complete case analysis, and (iii)
multiple imputation. Performance was assessed in terms of bias in estimating
exposure-outcome effects \add{among the exposed}, standard error, mean squared
error and coverage. Applying the CART algorithm directly to incomplete data
resulted in bias, even in scenarios where data were missing completely at
random. Overall, multiple imputation followed by CART resulted in the best
performance. Our study showed that automatic handling of missing data in CART
can cause serious bias and does not outperform multiple imputation as a means
to account for missing data.
|
Supervised and Semi-Supervised Deep Neural Networks for CSI-Based
Authentication
|
From the viewpoint of physical-layer authentication, spoofing attacks can be
foiled by checking channel state information (CSI). Existing CSI-based
authentication algorithms mostly require a deep knowledge of the channel to
deliver decent performance. In this paper, we investigate CSI-based
authenticators that can spare the effort to predetermine channel properties by
utilizing deep neural networks (DNNs). We first propose a convolutional neural
network (CNN)-enabled authenticator that is able to extract the local features
in CSI. Next, we employ the recurrent neural network (RNN) to capture the
dependencies between different frequencies in CSI. In addition, we propose to
use the convolutional recurrent neural network (CRNN)---a combination of the
CNN and the RNN---to learn local and contextual information in CSI for user
authentication. To effectively train these DNNs, one needs a large amount of
labeled channel records. However, it is often expensive to label large channel
observations in the presence of a spoofer. In view of this, we further study a
case in which only a small part of the the channel observations are labeled. To
handle it, we extend these DNNs-enabled approaches into semi-supervised ones.
This extension is based on a semi-supervised learning technique that employs
both the labeled and unlabeled data to train a DNN. To be specific, our
semi-supervised method begins by generating pseudo labels for the unlabeled
channel samples through implementing the K-means algorithm in a semi-supervised
manner. Subsequently, both the labeled and pseudo labeled data are exploited to
pre-train a DNN, which is then fine-tuned based on the labeled channel records.
|
How good is my GAN?
|
Generative adversarial networks (GANs) are one of the most popular methods
for generating images today. While impressive results have been validated by
visual inspection, a number of quantitative criteria have emerged only
recently. We argue here that the existing ones are insufficient and need to be
in adequation with the task at hand. In this paper we introduce two measures
based on image classification---GAN-train and GAN-test, which approximate the
recall (diversity) and precision (quality of the image) of GANs respectively.
We evaluate a number of recent GAN approaches based on these two measures and
demonstrate a clear difference in performance. Furthermore, we observe that the
increasing difficulty of the dataset, from CIFAR10 over CIFAR100 to ImageNet,
shows an inverse correlation with the quality of the GANs, as clearly evident
from our measures.
|
Pre-trainable Reservoir Computing with Recursive Neural Gas
|
Echo State Networks (ESN) are a class of Recurrent Neural Networks (RNN) that
has gained substantial popularity due to their effectiveness, ease of use and
potential for compact hardware implementation. An ESN contains the three
network layers input, reservoir and readout where the reservoir is the truly
recurrent network. The input and reservoir layers of an ESN are initialized at
random and never trained afterwards and the training of the ESN is applied to
the readout layer only. The alternative of Recursive Neural Gas (RNG) is one of
the many proposals of fully-trainable reservoirs that can be found in the
literature. Although some improvements in performance have been reported with
RNG, to the best of authors' knowledge, no experimental comparative results are
known with benchmarks for which ESN is known to yield excellent results. This
work describes an accurate model of RNG together with some extensions to the
models presented in the literature and shows comparative results on three
well-known and accepted datasets. The experimental results obtained show that,
under specific circumstances, RNG-based reservoirs can achieve better
performance.
|
Backprop-Q: Generalized Backpropagation for Stochastic Computation
Graphs
|
In real-world scenarios, it is appealing to learn a model carrying out
stochastic operations internally, known as stochastic computation graphs
(SCGs), rather than learning a deterministic mapping. However, standard
backpropagation is not applicable to SCGs. We attempt to address this issue
from the angle of cost propagation, with local surrogate costs, called
Q-functions, constructed and learned for each stochastic node in an SCG. Then,
the SCG can be trained based on these surrogate costs using standard
backpropagation. We propose the entire framework as a solution to generalize
backpropagation for SCGs, which resembles an actor-critic architecture but
based on a graph. For broad applicability, we study a variety of SCG structures
from one cost to multiple costs. We utilize recent advances in reinforcement
learning (RL) and variational Bayes (VB), such as off-policy critic learning
and unbiased-and-low-variance gradient estimation, and review them in the
context of SCGs. The generalized backpropagation extends transported learning
signals beyond gradients between stochastic nodes while preserving the benefit
of backpropagating gradients through deterministic nodes. Experimental
suggestions and concerns are listed to help design and test any specific model
using this framework.
|
A machine learning framework for data driven acceleration of
computations of differential equations
|
We propose a machine learning framework to accelerate numerical computations
of time-dependent ODEs and PDEs. Our method is based on recasting
(generalizations of) existing numerical methods as artificial neural networks,
with a set of trainable parameters. These parameters are determined in an
offline training process by (approximately) minimizing suitable (possibly
non-convex) loss functions by (stochastic) gradient descent methods. The
proposed algorithm is designed to be always consistent with the underlying
differential equation. Numerical experiments involving both linear and
non-linear ODE and PDE model problems demonstrate a significant gain in
computational efficiency over standard numerical methods.
|
Conditional Information Gain Networks
|
Deep neural network models owe their representational power to the high
number of learnable parameters. It is often infeasible to run these largely
parametrized deep models in limited resource environments, like mobile phones.
Network models employing conditional computing are able to reduce computational
requirements while achieving high representational power, with their ability to
model hierarchies. We propose Conditional Information Gain Networks, which
allow the feed forward deep neural networks to execute conditionally, skipping
parts of the model based on the sample and the decision mechanisms inserted in
the architecture. These decision mechanisms are trained using cost functions
based on differentiable Information Gain, inspired by the training procedures
of decision trees. These information gain based decision mechanisms are
differentiable and can be trained end-to-end using a unified framework with a
general cost function, covering both classification and decision losses. We
test the effectiveness of the proposed method on MNIST and recently introduced
Fashion MNIST datasets and show that our information gain based conditional
execution approach can achieve better or comparable classification results
using significantly fewer parameters, compared to standard convolutional neural
network baselines.
|
Enhancing keyword correlation for event detection in social networks
using SVD and k-means: Twitter case study
|
Extracting textual features from tweets is a challenging process due to the
noisy nature of the content and the weak signal of most of the words used. In
this paper, we propose using singular value decomposition (SVD) with clustering
to enhance the signals of the textual features in the tweets to improve the
correlation with events. The proposed technique applies SVD to the time series
vector for each feature to factorize the matrix of feature/day counts, in order
to ensure the independence of the feature vectors. Afterwards, the k-means
clustering is applied to build a look-up table that maps members of each
cluster to the cluster-centroid. The lookup table is used to map each feature
in the original data to the centroid of its cluster, then we calculate the sum
of the term frequency vectors of all features in each cluster to the
term-frequency-vector of the cluster centroid. To test the technique we
calculated the correlations of the cluster centroids with the golden standard
record (GSR) vector before and after summing the vectors of the cluster members
to the centroid-vector. The proposed method is applied to multiple correlation
techniques including the Pearson, Spearman, distance correlation and Kendal
Tao. The experiments have also considered the different word forms and lengths
of the features including keywords, n-grams, skip-grams and bags-of-words. The
correlation results are enhanced significantly as the highest correlation
scores have increased from 0.3 to 0.6, and the average correlation scores have
increased from 0.3 to 0.4.
|
Deep Learning Detection Networks in MIMO Decode-Forward Relay Channels
|
In this paper, we consider signal detection algorithms in a multiple-input
multiple-output (MIMO) decode-forward (DF) relay channel with one source, one
relay, and one destination. The existing suboptimal near maximum likelihood
(NML) detector and the NML with two-level pair-wise error probability
(NMLw2PEP) detector achieve excellent performance with instantaneous channel
state information (CSI) of the source-relay (SR) link and with statistical CSI
of the SR link, respectively. However, the NML detectors require an
exponentially increasing complexity as the number of transmit antennas
increases. Using deep learning algorithms, NML-based detection networks
(NMLDNs) are proposed with and without the CSI of the SR link at the
destination. The NMLDNs detect signals in changing channels after a single
training using a large number of randomly distributed channels. The detection
networks require much lower detection complexity than the exhaustive search NML
detectors while exhibiting good performance. To evaluate the performance, we
introduce semidefinite relaxation detectors with polynomial complexity based on
the NML detectors. Additionally, new linear detectors based on the zero
gradient of the NML metrics are proposed. Applying various detection algorithms
at the relay (DetR) and detection algorithms at the destination (DetD), we
present some DetR-DetD methods in MIMO DF relay channels. An appropriate
DetR-DetD method can be employed according to the required error probability
and detection complexity. The complexity analysis and simulation results
validate the arguments of this paper.
|
Redundancy Coefficient Gradual Up-weighting-based Mutual Information
Feature Selection Technique for Crypto-ransomware Early Detection
|
Crypto-ransomware is characterized by its irreversible effect even after the
detection and removal. As such, the early detection is crucial to protect user
data and files of being held to ransom. Several solutions have proposed
utilizing the data extracted during the initial phases of the attacks before
the encryption takes place. However, the lack of enough data at the early
phases of the attack along with high dimensional features space renders the
model prone to overfitting which decreases its detection accuracy. To this end,
this paper proposed a novel redundancy coefficient gradual up-weighting
approach that was incorporated to the calculation of redundancy term of mutual
information to improve the feature selection process and enhance the accuracy
of the detection model. Several machine learning classifiers were used to
evaluate the detection performance of the proposed techniques. The experimental
results show that the accuracy of proposed techniques achieved higher detection
accuracy. Those results demonstrate the efficacy of the proposed techniques for
the early detection tasks.
|
Perturb and Combine to Identify Influential Spreaders in Real-World
Networks
|
Some of the most effective influential spreader detection algorithms are
unstable to small perturbations of the network structure. Inspired by bagging
in Machine Learning, we propose the first Perturb and Combine (P&C) procedure
for networks. It (1) creates many perturbed versions of a given graph, (2)
applies a node scoring function separately to each graph, and (3) combines the
results. Experiments conducted on real-world networks of various sizes with the
k-core, generalized k-core, and PageRank algorithms reveal that P&C brings
substantial improvements. Moreover, this performance boost can be obtained at
almost no extra cost through parallelization. Finally, a bias-variance analysis
suggests that P&C works mainly by reducing bias, and that therefore, it should
be capable of improving the performance of all vertex scoring functions,
including stable ones.
|
Contextual Stochastic Block Models
|
We provide the first information theoretic tight analysis for inference of
latent community structure given a sparse graph along with high dimensional
node covariates, correlated with the same latent communities. Our work bridges
recent theoretical breakthroughs in the detection of latent community structure
without nodes covariates and a large body of empirical work using diverse
heuristics for combining node covariates with graphs for inference. The
tightness of our analysis implies in particular, the information theoretical
necessity of combining the different sources of information. Our analysis holds
for networks of large degrees as well as for a Gaussian version of the model.
|
Acoustic-to-Word Recognition with Sequence-to-Sequence Models
|
Acoustic-to-Word recognition provides a straightforward solution to
end-to-end speech recognition without needing external decoding, language model
re-scoring or lexicon. While character-based models offer a natural solution to
the out-of-vocabulary problem, word models can be simpler to decode and may
also be able to directly recognize semantically meaningful units. We present
effective methods to train Sequence-to-Sequence models for direct word-level
recognition (and character-level recognition) and show an absolute improvement
of 4.4-5.0\% in Word Error Rate on the Switchboard corpus compared to prior
work. In addition to these promising results, word-based models are more
interpretable than character models, which have to be composed into words using
a separate decoding step. We analyze the encoder hidden states and the
attention behavior, and show that location-aware attention naturally represents
words as a single speech-word-vector, despite spanning multiple frames in the
input. We finally show that the Acoustic-to-Word model also learns to segment
speech into words with a mean standard deviation of 3 frames as compared with
human annotated forced-alignments for the Switchboard corpus.
|
Convolutional Neural Networks In Classifying Cancer Through DNA
Methylation
|
DNA Methylation has been the most extensively studied epigenetic mark.
Usually a change in the genotype, DNA sequence, leads to a change in the
phenotype, observable characteristics of the individual. But DNA methylation,
which happens in the context of CpG (cytosine and guanine bases linked by
phosphate backbone) dinucleotides, does not lead to a change in the original
DNA sequence but has the potential to change the phenotype. DNA methylation is
implicated in various biological processes and diseases including cancer. Hence
there is a strong interest in understanding the DNA methylation patterns across
various epigenetic related ailments in order to distinguish and diagnose the
type of disease in its early stages. In this work, the relationship between
methylated versus unmethylated CpG regions and cancer types is explored using
Convolutional Neural Networks (CNNs). A CNN based Deep Learning model that can
classify the cancer of a new DNA methylation profile based on the learning from
publicly available DNA methylation datasets is then proposed.
|
Repartitioning of the ComplexWebQuestions Dataset
|
Recently, Talmor and Berant (2018) introduced ComplexWebQuestions - a dataset
focused on answering complex questions by decomposing them into a sequence of
simpler questions and extracting the answer from retrieved web snippets. In
their work the authors used a pre-trained reading comprehension (RC) model
(Salant and Berant, 2018) to extract the answer from the web snippets. In this
short note we show that training a RC model directly on the training data of
ComplexWebQuestions reveals a leakage from the training set to the test set
that allows to obtain unreasonably high performance. As a solution, we
construct a new partitioning of ComplexWebQuestions that does not suffer from
this leakage and publicly release it. We also perform an empirical evaluation
on these two datasets and show that training a RC model on the training data
substantially improves state-of-the-art performance.
|
Finding Better Subword Segmentation for Neural Machine Translation
|
For different language pairs, word-level neural machine translation (NMT)
models with a fixed-size vocabulary suffer from the same problem of
representing out-of-vocabulary (OOV) words. The common practice usually
replaces all these rare or unknown words with a <UNK> token, which limits the
translation performance to some extent. Most of recent work handled such a
problem by splitting words into characters or other specially extracted subword
units to enable open-vocabulary translation. Byte pair encoding (BPE) is one of
the successful attempts that has been shown extremely competitive by providing
effective subword segmentation for NMT systems. In this paper, we extend the
BPE style segmentation to a general unsupervised framework with three
statistical measures: frequency (FRQ), accessor variety (AV) and description
length gain (DLG). We test our approach on two translation tasks: German to
English and Chinese to English. The experimental results show that AV and DLG
enhanced systems outperform the FRQ baseline in the frequency weighted schemes
at different significant levels.
|
Three hypergraph eigenvector centralities
|
Eigenvector centrality is a standard network analysis tool for determining
the importance of (or ranking of) entities in a connected system that is
represented by a graph. However, many complex systems and datasets have natural
multi-way interactions that are more faithfully modeled by a hypergraph. Here
we extend the notion of graph eigenvector centrality to uniform hypergraphs.
Traditional graph eigenvector centralities are given by a positive eigenvector
of the adjacency matrix, which is guaranteed to exist by the Perron-Frobenius
theorem under some mild conditions. The natural representation of a hypergraph
is a hypermatrix (colloquially, a tensor). Using recently established
Perron-Frobenius theory for tensors, we develop three tensor eigenvectors
centralities for hypergraphs, each with different interpretations. We show that
these centralities can reveal different information on real-world data by
analyzing hypergraphs constructed from n-gram frequencies, co-tagging on stack
exchange, and drug combinations observed in patient emergency room visits.
|
Variational Bayesian Reinforcement Learning with Regret Bounds
|
In reinforcement learning the Q-values summarize the expected future rewards
that the agent will attain. However, they cannot capture the epistemic
uncertainty about those rewards. In this work we derive a new Bellman operator
with associated fixed point we call the `knowledge values'. These K-values
compress both the expected future rewards and the epistemic uncertainty into a
single value, so that high uncertainty, high reward, or both, can yield high
K-values. The key principle is to endow the agent with a risk-seeking utility
function that is carefully tuned to balance exploration and exploitation. When
the agent follows a Boltzmann policy over the K-values it yields a Bayes regret
bound of $\tilde O(L^{3/2} \sqrt{S A T})$, where $L$ is the time horizon, $S$
is the number of states, $A$ is the number of actions, and $T$ is the total
number of elapsed timesteps. We show deep connections of this approach to the
soft-max and maximum-entropy strands of research in reinforcement learning.
|
A Surprising Linear Relationship Predicts Test Performance in Deep
Networks
|
Given two networks with the same training loss on a dataset, when would they
have drastically different test losses and errors? Better understanding of this
question of generalization may improve practical applications of deep networks.
In this paper we show that with cross-entropy loss it is surprisingly simple to
induce significantly different generalization performances for two networks
that have the same architecture, the same meta parameters and the same training
error: one can either pretrain the networks with different levels of
"corrupted" data or simply initialize the networks with weights of different
Gaussian standard deviations. A corollary of recent theoretical results on
overfitting shows that these effects are due to an intrinsic problem of
measuring test performance with a cross-entropy/exponential-type loss, which
can be decomposed into two components both minimized by SGD -- one of which is
not related to expected classification performance. However, if we factor out
this component of the loss, a linear relationship emerges between training and
test losses. Under this transformation, classical generalization bounds are
surprisingly tight: the empirical/training loss is very close to the
expected/test loss. Furthermore, the empirical relation between classification
error and normalized cross-entropy loss seem to be approximately monotonic
|
Limitations of the Lipschitz constant as a defense against adversarial
examples
|
Several recent papers have discussed utilizing Lipschitz constants to limit
the susceptibility of neural networks to adversarial examples. We analyze
recently proposed methods for computing the Lipschitz constant. We show that
the Lipschitz constant may indeed enable adversarially robust neural networks.
However, the methods currently employed for computing it suffer from
theoretical and practical limitations. We argue that addressing this
shortcoming is a promising direction for future research into certified
adversarial defenses.
|
Convergence Rates of Gaussian ODE Filters
|
A recently-introduced class of probabilistic (uncertainty-aware) solvers for
ordinary differential equations (ODEs) applies Gaussian (Kalman) filtering to
initial value problems. These methods model the true solution $x$ and its first
$q$ derivatives \emph{a priori} as a Gauss--Markov process $\boldsymbol{X}$,
which is then iteratively conditioned on information about $\dot{x}$. This
article establishes worst-case local convergence rates of order $q+1$ for a
wide range of versions of this Gaussian ODE filter, as well as global
convergence rates of order $q$ in the case of $q=1$ and an integrated Brownian
motion prior, and analyses how inaccurate information on $\dot{x}$ coming from
approximate evaluations of $f$ affects these rates. Moreover, we show that, in
the globally convergent case, the posterior credible intervals are well
calibrated in the sense that they globally contract at the same rate as the
truncation error. We illustrate these theoretical results by numerical
experiments which might indicate their generalizability to $q \in
\{2,3,\dots\}$.
|
PADME: A Deep Learning-based Framework for Drug-Target Interaction
Prediction
|
In silico drug-target interaction (DTI) prediction is an important and
challenging problem in biomedical research with a huge potential benefit to the
pharmaceutical industry and patients. Most existing methods for DTI prediction
including deep learning models generally have binary endpoints, which could be
an oversimplification of the problem, and those methods are typically unable to
handle cold-target problems, i.e., problems involving target protein that never
appeared in the training set. Towards this, we contrived PADME (Protein And
Drug Molecule interaction prEdiction), a framework based on Deep Neural
Networks, to predict real-valued interaction strength between compounds and
proteins without requiring feature engineering. PADME takes both compound and
protein information as inputs, so it is capable of solving cold-target (and
cold-drug) problems. To our knowledge, we are the first to combine Molecular
Graph Convolution (MGC) for compound featurization with protein descriptors for
DTI prediction. We used multiple cross-validation split schemes and evaluation
metrics to measure the performance of PADME on multiple datasets, including the
ToxCast dataset, and PADME consistently dominates baseline methods. The results
of a case study, which predicts the binding affinity between various compounds
and androgen receptor (AR), suggest PADME's potential in drug development. The
scalability of PADME is another advantage in the age of Big Data.
|
Multi-Perspective Neural Architecture for Recommendation System
|
Currently, there starts a research trend to leverage neural architecture for
recommendation systems. Though several deep recommender models are proposed,
most methods are too simple to characterize users' complex preference. In this
paper, for a fine-grain analysis, users' ratings are explained from multiple
perspectives, based on which, we propose our neural architecture. Specifically,
our model employs several sequential stages to encode the user and item into
hidden representations. In one stage, the user and item are represented from
multiple perspectives and in each perspective, the representations of user and
item put attentions to each other. Last, we metric the output representations
of final stage to approach the users' rating. Extensive experiments demonstrate
that our method achieves substantial improvements against baselines.
|
Stripe-Based Fragility Analysis of Concrete Bridge Classes Using Machine
Learning Techniques
|
A framework for the generation of bridge-specific fragility utilizing the
capabilities of machine learning and stripe-based approach is presented in this
paper. The proposed methodology using random forests helps to generate or
update fragility curves for a new set of input parameters with less
computational effort and expensive re-simulation. The methodology does not
place any assumptions on the demand model of various components and helps to
identify the relative importance of each uncertain variable in their seismic
demand model. The methodology is demonstrated through the case studies of
multi-span concrete bridges in California. Geometric, material and structural
uncertainties are accounted for in the generation of bridge models and
fragility curves. It is also noted that the traditional lognormality assumption
on the demand model leads to unrealistic fragility estimates. Fragility results
obtained the proposed methodology curves can be deployed in risk assessment
platform such as HAZUS for regional loss estimation.
|
Deep Contextual Multi-armed Bandits
|
Contextual multi-armed bandit problems arise frequently in important
industrial applications. Existing solutions model the context either linearly,
which enables uncertainty driven (principled) exploration, or non-linearly, by
using epsilon-greedy exploration policies. Here we present a deep learning
framework for contextual multi-armed bandits that is both non-linear and
enables principled exploration at the same time. We tackle the exploration vs.
exploitation trade-off through Thompson sampling by exploiting the connection
between inference time dropout and sampling from the posterior over the weights
of a Bayesian neural network. In order to adjust the level of exploration
automatically as more data is made available to the model, the dropout rate is
learned rather than considered a hyperparameter. We demonstrate that our
approach substantially reduces regret on two tasks (the UCI Mushroom task and
the Casino Parity task) when compared to 1) non-contextual bandits, 2)
epsilon-greedy deep contextual bandits, and 3) fixed dropout rate deep
contextual bandits. Our approach is currently being applied to marketing
optimization problems at HubSpot.
|
Coreset-Based Neural Network Compression
|
We propose a novel Convolutional Neural Network (CNN) compression algorithm
based on coreset representations of filters. We exploit the redundancies extant
in the space of CNN weights and neuronal activations (across samples) in order
to obtain compression. Our method requires no retraining, is easy to implement,
and obtains state-of-the-art compression performance across a wide variety of
CNN architectures. Coupled with quantization and Huffman coding, we create
networks that provide AlexNet-like accuracy, with a memory footprint that is
$832\times$ smaller than the original AlexNet, while also introducing
significant reductions in inference time as well. Additionally these compressed
networks when fine-tuned, successfully generalize to other domains as well.
|
Binacox: automatic cut-point detection in high-dimensional Cox model
with applications in genetics
|
We introduce the binacox, a prognostic method to deal with the problem of
detecting multiple cut-points per features in a multivariate setting where a
large number of continuous features are available. The method is based on the
Cox model and combines one-hot encoding with the binarsity penalty, which uses
total-variation regularization together with an extra linear constraint, and
enables feature selection. Original nonasymptotic oracle inequalities for
prediction (in terms of Kullback-Leibler divergence) and estimation with a fast
rate of convergence are established. The statistical performance of the method
is examined in an extensive Monte Carlo simulation study, and then illustrated
on three publicly available genetic cancer datasets. On these high-dimensional
datasets, our proposed method significantly outperforms state-of-the-art
survival models regarding risk prediction in terms of the C-index, with a
computing time orders of magnitude faster. In addition, it provides powerful
interpretability from a clinical perspective by automatically pinpointing
significant cut-points in relevant variables.
|
Comparison of methods for early-readmission prediction in a
high-dimensional heterogeneous covariates and time-to-event outcome framework
|
Background: Choosing the most performing method in terms of outcome
prediction or variables selection is a recurring problem in prognosis studies,
leading to many publications on methods comparison. But some aspects have
received little attention. First, most comparison studies treat prediction
performance and variable selection aspects separately. Second, methods are
either compared within a binary outcome setting (based on an arbitrarily chosen
delay) or within a survival setting, but not both. In this paper, we propose a
comparison methodology to weight up those different settings both in terms of
prediction and variables selection, while incorporating advanced machine
learning strategies. Methods: Using a high-dimensional case study on a
sickle-cell disease (SCD) cohort, we compare 8 statistical methods. In the
binary outcome setting, we consider logistic regression (LR), support vector
machine (SVM), random forest (RF), gradient boosting (GB) and neural network
(NN); while on the survival analysis setting, we consider the Cox Proportional
Hazards (PH), the CURE and the C-mix models. We then compare performances of
all methods both in terms of risk prediction and variable selection, with a
focus on the use of Elastic-Net regularization technique. Results: Among all
assessed statistical methods assessed, the C-mix model yields the better
performances in both the two considered settings, as well as interesting
interpretation aspects. There is some consistency in selected covariates across
methods within a setting, but not much across the two settings. Conclusions: It
appears that learning withing the survival setting first, and then going back
to a binary prediction using the survival estimates significantly enhance
binary predictions.
|
Iterative evaluation of LSTM cells
|
In this work we present a modification in the conventional flow of
information through a LSTM network, which we consider well suited for RNNs in
general. The modification leads to a iterative scheme where the computations
performed by the LSTM cell are repeated over a constant input and cell state
values, while updating the hidden state a finite number of times. We provide
theoretical and empirical evidence to support the augmented capabilities of the
iterative scheme and show examples related to language modeling. The
modification yields an enhancement in the model performance comparable with the
original model augmented more than 3 times in terms of the total amount of
parameters.
|
Applying Domain Randomization to Synthetic Data for Object Category
Detection
|
Recent advances in deep learning-based object detection techniques have
revolutionized their applicability in several fields. However, since these
methods rely on unwieldy and large amounts of data, a common practice is to
download models pre-trained on standard datasets and fine-tune them for
specific application domains with a small set of domain relevant images. In
this work, we show that using synthetic datasets that are not necessarily
photo-realistic can be a better alternative to simply fine-tune pre-trained
networks. Specifically, our results show an impressive 25% improvement in the
mAP metric over a fine-tuning baseline when only about 200 labelled images are
available to train. Finally, an ablation study of our results is presented to
delineate the individual contribution of different components in the
randomization pipeline.
|
Understanding and representing the semantics of large structured
documents
|
Understanding large, structured documents like scholarly articles, requests
for proposals or business reports is a complex and difficult task. It involves
discovering a document's overall purpose and subject(s), understanding the
function and meaning of its sections and subsections, and extracting low level
entities and facts about them. In this research, we present a deep learning
based document ontology to capture the general purpose semantic structure and
domain specific semantic concepts from a large number of academic articles and
business documents. The ontology is able to describe different functional parts
of a document, which can be used to enhance semantic indexing for a better
understanding by human beings and machines. We evaluate our models through
extensive experiments on datasets of scholarly articles from arXiv and Request
for Proposal documents.
|
Modular Mechanistic Networks: On Bridging Mechanistic and
Phenomenological Models with Deep Neural Networks in Natural Language
Processing
|
Natural language processing (NLP) can be done using either top-down (theory
driven) and bottom-up (data driven) approaches, which we call mechanistic and
phenomenological respectively. The approaches are frequently considered to
stand in opposition to each other. Examining some recent approaches in deep
learning we argue that deep neural networks incorporate both perspectives and,
furthermore, that leveraging this aspect of deep learning may help in solving
complex problems within language technology, such as modelling language and
perception in the domain of spatial cognition.
|
Predicting Acute Kidney Injury at Hospital Re-entry Using
High-dimensional Electronic Health Record Data
|
Acute Kidney Injury (AKI), a sudden decline in kidney function, is associated
with increased mortality, morbidity, length of stay, and hospital cost. Since
AKI is sometimes preventable, there is great interest in prediction. Most
existing studies consider all patients and therefore restrict to features
available in the first hours of hospitalization. Here, the focus is instead on
rehospitalized patients, a cohort in which rich longitudinal features from
prior hospitalizations can be analyzed. Our objective is to provide a risk
score directly at hospital re-entry. Gradient boosting, penalized logistic
regression (with and without stability selection), and a recurrent neural
network are trained on two years of adult inpatient EHR data (3,387 attributes
for 34,505 patients who generated 90,013 training samples with 5,618 cases and
84,395 controls). Predictions are internally evaluated with 50 iterations of
5-fold grouped cross-validation with special emphasis on calibration, an
analysis of which is performed at the patient as well as hospitalization level.
Error is assessed with respect to diagnosis, race, age, gender, AKI
identification method, and hospital utilization. In an additional experiment,
the regularization penalty is severely increased to induce parsimony and
interpretability. Predictors identified for rehospitalized patients are also
reported with a special analysis of medications that might be modifiable risk
factors. Insights from this study might be used to construct a predictive tool
for AKI in rehospitalized patients. An accurate estimate of AKI risk at
hospital entry might serve as a prior for an admitting provider or another
predictive algorithm.
|
Do Better ImageNet Models Transfer Better... for Image Recommendation?
|
Visual embeddings from Convolutional Neural Networks (CNN) trained on the
ImageNet dataset for the ILSVRC challenge have shown consistently good
performance for transfer learning and are widely used in several tasks,
including image recommendation. However, some important questions have not yet
been answered in order to use these embeddings for a larger scope of
recommendation domains: a) Do CNNs that perform better in ImageNet are also
better for transfer learning in content-based image recommendation?, b) Does
fine-tuning help to improve performance? and c) Which is the best way to
perform the fine-tuning?
In this paper we compare several CNN models pre-trained with ImageNet to
evaluate their transfer learning performance to an artwork image recommendation
task. Our results indicate that models with better performance in the ImageNet
challenge do not always imply better transfer learning for recommendation tasks
(e.g. NASNet vs. ResNet). Our results also show that fine-tuning can be helpful
even with a small dataset, but not every fine-tuning works. Our results can
inform other researchers and practitioners on how to train their CNNs for
better transfer learning towards image recommendation systems.
|
Differentiable Perturb-and-Parse: Semi-Supervised Parsing with a
Structured Variational Autoencoder
|
Human annotation for syntactic parsing is expensive, and large resources are
available only for a fraction of languages. A question we ask is whether one
can leverage abundant unlabeled texts to improve syntactic parsers, beyond just
using the texts to obtain more generalisable lexical features (i.e. beyond word
embeddings). To this end, we propose a novel latent-variable generative model
for semi-supervised syntactic dependency parsing. As exact inference is
intractable, we introduce a differentiable relaxation to obtain approximate
samples and compute gradients with respect to the parser parameters. Our method
(Differentiable Perturb-and-Parse) relies on differentiable dynamic programming
over stochastically perturbed edge scores. We demonstrate effectiveness of our
approach with experiments on English, French and Swedish.
|
Neural State Classification for Hybrid Systems
|
We introduce the State Classification Problem (SCP) for hybrid systems, and
present Neural State Classification (NSC) as an efficient solution technique.
SCP generalizes the model checking problem as it entails classifying each state
$s$ of a hybrid automaton as either positive or negative, depending on whether
or not $s$ satisfies a given time-bounded reachability specification. This is
an interesting problem in its own right, which NSC solves using
machine-learning techniques, Deep Neural Networks in particular. State
classifiers produced by NSC tend to be very efficient (run in constant time and
space), but may be subject to classification errors. To quantify and mitigate
such errors, our approach comprises: i) techniques for certifying, with
statistical guarantees, that an NSC classifier meets given accuracy levels; ii)
tuning techniques, including a novel technique based on adversarial sampling,
that can virtually eliminate false negatives (positive states classified as
negative), thereby making the classifier more conservative. We have applied NSC
to six nonlinear hybrid system benchmarks, achieving an accuracy of 99.25% to
99.98%, and a false-negative rate of 0.0033 to 0, which we further reduced to
0.0015 to 0 after tuning the classifier. We believe that this level of accuracy
is acceptable in many practical applications, and that these results
demonstrate the promise of the NSC approach.
|
General-purpose Tagging of Freesound Audio with AudioSet Labels: Task
Description, Dataset, and Baseline
|
This paper describes Task 2 of the DCASE 2018 Challenge, titled
"General-purpose audio tagging of Freesound content with AudioSet labels". This
task was hosted on the Kaggle platform as "Freesound General-Purpose Audio
Tagging Challenge". The goal of the task is to build an audio tagging system
that can recognize the category of an audio clip from a subset of 41 diverse
categories drawn from the AudioSet Ontology. We present the task, the dataset
prepared for the competition, and a baseline system.
|
A Data-Efficient Approach to Precise and Controlled Pushing
|
Decades of research in control theory have shown that simple controllers,
when provided with timely feedback, can control complex systems. Pushing is an
example of a complex mechanical system that is difficult to model accurately
due to unknown system parameters such as coefficients of friction and pressure
distributions. In this paper, we explore the data-complexity required for
controlling, rather than modeling, such a system. Results show that a
model-based control approach, where the dynamical model is learned from data,
is capable of performing complex pushing trajectories with a minimal amount of
training data (10 data points). The dynamics of pushing interactions are
modeled using a Gaussian process (GP) and are leveraged within a model
predictive control approach that linearizes the GP and imposes actuator and
task constraints for a planar manipulation task.
|
Meta-learning autoencoders for few-shot prediction
|
Compared to humans, machine learning models generally require significantly
more training examples and fail to extrapolate from experience to solve
previously unseen challenges. To help close this performance gap, we augment
single-task neural networks with a meta-recognition model which learns a
succinct model code via its autoencoder structure, using just a few informative
examples. The model code is then employed by a meta-generative model to
construct parameters for the task-specific model. We demonstrate that for
previously unseen tasks, without additional training, this Meta-Learning
Autoencoder (MeLA) framework can build models that closely match the true
underlying models, with loss significantly lower than given by fine-tuned
baseline networks, and performance that compares favorably with
state-of-the-art meta-learning algorithms. MeLA also adds the ability to
identify influential training examples and predict which additional data will
be most valuable to acquire to improve model prediction.
|
EBIC: an open source software for high-dimensional and big data
biclustering analyses
|
Motivation: In this paper we present the latest release of EBIC, a
next-generation biclustering algorithm for mining genetic data. The major
contribution of this paper is adding support for big data, making it possible
to efficiently run large genomic data mining analyses. Additional enhancements
include integration with R and Bioconductor and an option to remove influence
of missing value on the final result.
Results: EBIC was applied to datasets of different sizes, including a large
DNA methylation dataset with 436,444 rows. For the largest dataset we observed
over 6.6 fold speedup in computation time on a cluster of 8 GPUs compared to
running the method on a single GPU. This proves high scalability of the
algorithm.
Availability: The latest version of EBIC could be downloaded from
http://github.com/EpistasisLab/ebic . Installation and usage instructions are
also available online.
|
Multi-Agent Generative Adversarial Imitation Learning
|
Imitation learning algorithms can be used to learn a policy from expert
demonstrations without access to a reward signal. However, most existing
approaches are not applicable in multi-agent settings due to the existence of
multiple (Nash) equilibria and non-stationary environments. We propose a new
framework for multi-agent imitation learning for general Markov games, where we
build upon a generalized notion of inverse reinforcement learning. We further
introduce a practical multi-agent actor-critic algorithm with good empirical
performance. Our method can be used to imitate complex behaviors in
high-dimensional environments with multiple cooperative or competing agents.
|
HiDDeN: Hiding Data With Deep Networks
|
Recent work has shown that deep neural networks are highly sensitive to tiny
perturbations of input images, giving rise to adversarial examples. Though this
property is usually considered a weakness of learned models, we explore whether
it can be beneficial. We find that neural networks can learn to use invisible
perturbations to encode a rich amount of useful information. In fact, one can
exploit this capability for the task of data hiding. We jointly train encoder
and decoder networks, where given an input message and cover image, the encoder
produces a visually indistinguishable encoded image, from which the decoder can
recover the original message. We show that these encodings are competitive with
existing data hiding algorithms, and further that they can be made robust to
noise: our models learn to reconstruct hidden information in an encoded image
despite the presence of Gaussian blurring, pixel-wise dropout, cropping, and
JPEG compression. Even though JPEG is non-differentiable, we show that a robust
model can be trained using differentiable approximations. Finally, we
demonstrate that adversarial training improves the visual quality of encoded
images.
|
Computationally Efficient Measures of Internal Neuron Importance
|
The challenge of assigning importance to individual neurons in a network is
of interest when interpreting deep learning models. In recent work, Dhamdhere
et al. proposed Total Conductance, a "natural refinement of Integrated
Gradients" for attributing importance to internal neurons. Unfortunately, the
authors found that calculating conductance in tensorflow required the addition
of several custom gradient operators and did not scale well. In this work, we
show that the formula for Total Conductance is mathematically equivalent to
Path Integrated Gradients computed on a hidden layer in the network. We provide
a scalable implementation of Total Conductance using standard tensorflow
gradient operators that we call Neuron Integrated Gradients. We compare Neuron
Integrated Gradients to DeepLIFT, a pre-existing computationally efficient
approach that is applicable to calculating internal neuron importance. We find
that DeepLIFT produces strong empirical results and is faster to compute, but
because it lacks the theoretical properties of Neuron Integrated Gradients, it
may not always be preferred in practice. Colab notebook reproducing results:
http://bit.ly/neuronintegratedgradients
|
Rethinking the Form of Latent States in Image Captioning
|
RNNs and their variants have been widely adopted for image captioning. In
RNNs, the production of a caption is driven by a sequence of latent states.
Existing captioning models usually represent latent states as vectors, taking
this practice for granted. We rethink this choice and study an alternative
formulation, namely using two-dimensional maps to encode latent states. This is
motivated by the curiosity about a question: how the spatial structures in the
latent states affect the resultant captions? Our study on MSCOCO and Flickr30k
leads to two significant observations. First, the formulation with 2D states is
generally more effective in captioning, consistently achieving higher
performance with comparable parameter sizes. Second, 2D states preserve spatial
locality. Taking advantage of this, we visually reveal the internal dynamics in
the process of caption generation, as well as the connections between input
visual domain and output linguistic domain.
|
Learning to guide task and motion planning using score-space
representation
|
In this paper, we propose a learning algorithm that speeds up the search in
task and motion planning problems. Our algorithm proposes solutions to three
different challenges that arise in learning to improve planning efficiency:
what to predict, how to represent a planning problem instance, and how to
transfer knowledge from one problem instance to another. We propose a method
that predicts constraints on the search space based on a generic representation
of a planning problem instance, called score-space, where we represent a
problem instance in terms of the performance of a set of solutions attempted so
far. Using this representation, we transfer knowledge, in the form of
constraints, from previous problems based on the similarity in score space. We
design a sequential algorithm that efficiently predicts these constraints, and
evaluate it in three different challenging task and motion planning problems.
Results indicate that our approach performs orders of magnitudes faster than an
unguided planner
|
A Collaborative Approach to Angel and Venture Capital Investment
Recommendations
|
Matrix factorization was used to generate investment recommendations for
investors. An iterative conjugate gradient method was used to optimize the
regularized squared-error loss function. The number of latent factors, number
of iterations, and regularization values were explored. Overfitting can be
addressed by either early stopping or regularization parameter tuning. The
model achieved the highest average prediction accuracy of 13.3%. With a similar
model, the same dataset was used to generate investor recommendations for
companies undergoing fundraising, which achieved highest prediction accuracy of
11.1%.
|
Bayesian Optimal Design of Experiments For Inferring The Statistical
Expectation Of A Black-Box Function
|
Bayesian optimal design of experiments (BODE) has been successful in
acquiring information about a quantity of interest (QoI) which depends on a
black-box function. BODE is characterized by sequentially querying the function
at specific designs selected by an infill-sampling criterion. However, most
current BODE methods operate in specific contexts like optimization, or
learning a universal representation of the black-box function. The objective of
this paper is to design a BODE for estimating the statistical expectation of a
physical response surface. This QoI is omnipresent in uncertainty propagation
and design under uncertainty problems. Our hypothesis is that an optimal BODE
should be maximizing the expected information gain in the QoI. We represent the
information gain from a hypothetical experiment as the Kullback-Liebler (KL)
divergence between the prior and the posterior probability distributions of the
QoI. The prior distribution of the QoI is conditioned on the observed data and
the posterior distribution of the QoI is conditioned on the observed data and a
hypothetical experiment. The main contribution of this paper is the derivation
of a semi-analytic mathematical formula for the expected information gain about
the statistical expectation of a physical response. The developed BODE is
validated on synthetic functions with varying number of input-dimensions. We
demonstrate the performance of the methodology on a steel wire manufacturing
problem.
|
Automatic Detection of Node-Replication Attack in Vehicular Ad-hoc
Networks
|
Recent advances in smart cities applications enforce security threads such as
node replication attacks. Such attack is take place when the attacker plants a
replicated network node within the network. Vehicular Ad hoc networks are
connecting sensors that have limited resources and required the response time
to be as low as possible. In this type networks, traditional detection
algorithms of node replication attacks are not efficient. In this paper, we
propose an initial idea to apply a newly adapted statistical methodology that
can detect node replication attacks with high performance as compared to
state-of-the-art techniques. We provide a sufficient description of this
methodology and a road-map for testing and experiment its performance.
|
Towards a Deep Unified Framework for Nuclear Reactor Perturbation
Analysis
|
In this paper, we take the first steps towards a novel unified framework for
the analysis of perturbations in both the Time and Frequency domains. The
identification of type and source of such perturbations is fundamental for
monitoring reactor cores and guarantee safety while running at nominal
conditions. A 3D Convolutional Neural Network (3D-CNN) was employed to analyse
perturbations happening in the frequency domain, such as an absorber of
variable strength or propagating perturbation. Recurrent neural networks (RNN),
specifically Long Short-Term Memory (LSTM) networks were used to study signal
sequences related to perturbations induced in the time domain, including the
vibrations of fuel assemblies and the fluctuations of thermal-hydraulic
parameters at the inlet of the reactor coolant loops. 512 dimensional
representations were extracted from the 3D-CNN and LSTM architectures, and used
as input to a fused multi-sigmoid classification layer to recognise the
perturbation type. If the perturbation is in the frequency domain, a separate
fully-connected layer utilises said representations to regress the coordinates
of its source. The results showed that the perturbation type can be recognised
with high accuracy in all cases, and frequency domain scenario sources can be
localised with high precision.
|
ToriLLE: Learning Environment for Hand-to-Hand Combat
|
We present Toribash Learning Environment (ToriLLE), a learning environment
for machine learning agents based on the video game Toribash. Toribash is a
MuJoCo-like environment of two humanoid character fighting each other
hand-to-hand, controlled by changing actuation modes of the joints. Competitive
nature of Toribash as well its focused domain provide a platform for evaluating
self-play methods, and evaluating machine learning agents against human
players. In this paper we describe the environment with ToriLLE's capabilities
and limitations, and experimentally show its applicability as a learning
environment. The source code of the environment and conducted experiments can
be found at https://github.com/Miffyli/ToriLLE.
|
Effectiveness of Scaled Exponentially-Regularized Linear Units (SERLUs)
|
Recently, self-normalizing neural networks (SNNs) have been proposed with the
intention to avoid batch or weight normalization. The key step in SNNs is to
properly scale the exponential linear unit (referred to as SELU) to inherently
incorporate normalization based on central limit theory. SELU is a
monotonically increasing function, where it has an approximately constant
negative output for large negative input. In this work, we propose a new
activation function to break the monotonicity property of SELU while still
preserving the self-normalizing property. Differently from SELU, the new
function introduces a bump-shaped function in the region of negative input by
regularizing a linear function with a scaled exponential function, which is
referred to as a scaled exponentially-regularized linear unit (SERLU). The
bump-shaped function has approximately zero response to large negative input
while being able to push the output of SERLU towards zero mean statistically.
To effectively combat over-fitting, we develop a so-called shift-dropout for
SERLU, which includes standard dropout as a special case. Experimental results
on MNIST, CIFAR10 and CIFAR100 show that SERLU-based neural networks provide
consistently promising results in comparison to other 5 activation functions
including ELU, SELU, Swish, Leakly ReLU and ReLU.
|
A Unified Approximation Framework for Compressing and Accelerating Deep
Neural Networks
|
Deep neural networks (DNNs) have achieved significant success in a variety of
real world applications, i.e., image classification. However, tons of
parameters in the networks restrict the efficiency of neural networks due to
the large model size and the intensive computation. To address this issue,
various approximation techniques have been investigated, which seek for a light
weighted network with little performance degradation in exchange of smaller
model size or faster inference. Both low-rankness and sparsity are appealing
properties for the network approximation. In this paper we propose a unified
framework to compress the convolutional neural networks (CNNs) by combining
these two properties, while taking the nonlinear activation into consideration.
Each layer in the network is approximated by the sum of a structured sparse
component and a low-rank component, which is formulated as an optimization
problem. Then, an extended version of alternating direction method of
multipliers (ADMM) with guaranteed convergence is presented to solve the
relaxed optimization problem. Experiments are carried out on VGG-16, AlexNet
and GoogLeNet with large image classification datasets. The results outperform
previous work in terms of accuracy degradation, compression rate and speedup
ratio. The proposed method is able to remarkably compress the model (with up to
4.9x reduction of parameters) at a cost of little loss or without loss on
accuracy.
|
A Benchmark of Selected Algorithmic Differentiation Tools on Some
Problems in Computer Vision and Machine Learning
|
Algorithmic differentiation (AD) allows exact computation of derivatives
given only an implementation of an objective function. Although many AD tools
are available, a proper and efficient implementation of AD methods is not
straightforward. The existing tools are often too different to allow for a
general test suite. In this paper, we compare fifteen ways of computing
derivatives including eleven automatic differentiation tools implementing
various methods and written in various languages (C++, F#, MATLAB, Julia and
Python), two symbolic differentiation tools, finite differences, and
hand-derived computation.
We look at three objective functions from computer vision and machine
learning. These objectives are for the most part simple, in the sense that no
iterative loops are involved, and conditional statements are encapsulated in
functions such as {\tt abs} or {\tt logsumexp}. However, it is important for
the success of algorithmic differentiation that such `simple' objective
functions are handled efficiently, as so many problems in computer vision and
machine learning are of this form.
Of course, our results depend on programmer skill, and familiarity with the
tools. However, we contend that this paper presents an important datapoint: a
skilled programmer devoting roughly a week to each tool produced the timings we
present. We have made our implementations available as open source to allow the
community to replicate and update these benchmarks.
|
UNet++: A Nested U-Net Architecture for Medical Image Segmentation
|
In this paper, we present UNet++, a new, more powerful architecture for
medical image segmentation. Our architecture is essentially a deeply-supervised
encoder-decoder network where the encoder and decoder sub-networks are
connected through a series of nested, dense skip pathways. The re-designed skip
pathways aim at reducing the semantic gap between the feature maps of the
encoder and decoder sub-networks. We argue that the optimizer would deal with
an easier learning task when the feature maps from the decoder and encoder
networks are semantically similar. We have evaluated UNet++ in comparison with
U-Net and wide U-Net architectures across multiple medical image segmentation
tasks: nodule segmentation in the low-dose CT scans of chest, nuclei
segmentation in the microscopy images, liver segmentation in abdominal CT
scans, and polyp segmentation in colonoscopy videos. Our experiments
demonstrate that UNet++ with deep supervision achieves an average IoU gain of
3.9 and 3.4 points over U-Net and wide U-Net, respectively.
|
Rademacher Generalization Bounds for Classifier Chains
|
In this paper, we propose a new framework to study the generalization
property of classifier chains trained over observations associated with
multiple and interdependent class labels. The results are based on large
deviation inequalities for Lipschitz functions of weakly dependent sequences
proposed by Rio in 2000. We believe that the resulting generalization error
bound brings many advantages and could be adapted to other frameworks that
consider interdependent outputs. First, it explicitly exhibits the dependencies
between class labels. Secondly, it provides insights of the effect of the order
of the chain on the algorithm generalization performances. Finally, the two
dependency coefficients that appear in the bound could also be used to design
new strategies to decide the order of the chain.
|
Robust Tracking via Weighted Online Extreme Learning Machine
|
The tracking method based on the extreme learning machine (ELM) is efficient
and effective. ELM randomly generates input weights and biases in the hidden
layer, and then calculates and computes the output weights by reducing the
iterative solution to the problem of linear equations. Therefore, ELM offers
the satisfying classification performance and fast training time than other
discriminative models in tracking. However, the original ELM method often
suffers from the problem of the imbalanced classification distribution, which
is caused by few target objects, leading to under-fitting and more background
samples leading to over-fitting. Worse still, it reduces the robustness of
tracking under special conditions including occlusion, illumination, etc. To
address above problems, in this paper, we present a robust tracking algorithm.
First, we introduce the local weight matrix that is the dynamic creation from
the data distribution at the current frame in the original ELM so as to balance
between the empirical and structure risk, and fully learn the target object to
enhance the classification performance. Second, we improve it to the
incremental learning method ensuring tracking real-time and efficient. Finally,
the forgetting factor is used to strengthen the robustness for changing of the
classification distribution with time. Meanwhile, we propose a novel optimized
method to obtain the optimal sample as the target object, which avoids tracking
drift resulting from noisy samples. Therefore, our tracking method can fully
learn both of the target object and background information to enhance the
tracking performance, and it is evaluated in 20 challenge image sequences with
different attributes including illumination, occlusion, deformation, etc.,
which achieves better performance than several state-of-the-art methods in
terms of effectiveness and robustness.
|
DeepSPINE: Automated Lumbar Vertebral Segmentation, Disc-level
Designation, and Spinal Stenosis Grading Using Deep Learning
|
The high prevalence of spinal stenosis results in a large volume of MRI
imaging, yet interpretation can be time-consuming with high inter-reader
variability even among the most specialized radiologists. In this paper, we
develop an efficient methodology to leverage the subject-matter-expertise
stored in large-scale archival reporting and image data for a deep-learning
approach to fully-automated lumbar spinal stenosis grading. Specifically, we
introduce three major contributions: (1) a natural-language-processing scheme
to extract level-by-level ground-truth labels from free-text radiology reports
for the various types and grades of spinal stenosis (2) accurate vertebral
segmentation and disc-level localization using a U-Net architecture combined
with a spine-curve fitting method, and (3) a multi-input, multi-task, and
multi-class convolutional neural network to perform central canal and foraminal
stenosis grading on both axial and sagittal imaging series inputs with the
extracted report-derived labels applied to corresponding imaging level
segments. This study uses a large dataset of 22796 disc-levels extracted from
4075 patients. We achieve state-of-the-art performance on lumbar spinal
stenosis classification and expect the technique will increase both radiology
workflow efficiency and the perceived value of radiology reports for referring
clinicians and patients.
|
Medical Image Synthesis for Data Augmentation and Anonymization using
Generative Adversarial Networks
|
Data diversity is critical to success when training deep learning models.
Medical imaging data sets are often imbalanced as pathologic findings are
generally rare, which introduces significant challenges when training deep
learning models. In this work, we propose a method to generate synthetic
abnormal MRI images with brain tumors by training a generative adversarial
network using two publicly available data sets of brain MRI. We demonstrate two
unique benefits that the synthetic images provide. First, we illustrate
improved performance on tumor segmentation by leveraging the synthetic images
as a form of data augmentation. Second, we demonstrate the value of generative
models as an anonymization tool, achieving comparable tumor segmentation
results when trained on the synthetic data versus when trained on real subject
data. Together, these results offer a potential solution to two of the largest
challenges facing machine learning in medical imaging, namely the small
incidence of pathological findings, and the restrictions around sharing of
patient data.
|
Aggregated Learning: A Deep Learning Framework Based on
Information-Bottleneck Vector Quantization
|
Based on the notion of information bottleneck (IB), we formulate a
quantization problem called "IB quantization". We show that IB quantization is
equivalent to learning based on the IB principle. Under this equivalence, the
standard neural network models can be viewed as scalar (single sample) IB
quantizers. It is known, from conventional rate-distortion theory, that scalar
quantizers are inferior to vector (multi-sample) quantizers. Such a deficiency
then inspires us to develop a novel learning framework, AgrLearn, that
corresponds to vector IB quantizers for learning with neural networks. Unlike
standard networks, AgrLearn simultaneously optimizes against multiple data
samples. We experimentally verify that AgrLearn can result in significant
improvements when applied to several current deep learning architectures for
image recognition and text classification. We also empirically show that
AgrLearn can reduce up to 80% of the training samples needed for ResNet
training.
|
Novelty Detection Meets Collider Physics
|
Novelty detection is the machine learning task to recognize data, which
belong to an unknown pattern. Complementary to supervised learning, it allows
to analyze data model-independently. We demonstrate the potential role of
novelty detection in collider physics, using autoencoder-based deep neural
network. Explicitly, we develop a set of density-based novelty evaluators,
which are sensitive to the clustering of unknown-pattern testing data or
new-physics signal events, for the design of detection algorithms. We also
explore the influence of the known-pattern data fluctuations, arising from
non-signal regions, on detection sensitivity. Strategies to address it are
proposed. The algorithms are applied to detecting fermionic di-top partner and
resonant di-top productions at LHC, and exotic Higgs decays of two specific
modes at a $e^+e^-$ future collider. With parton-level analysis, we conclude
that potentially the new-physics benchmarks can be recognized with high
efficiency.
|
Seeded Graph Matching via Large Neighborhood Statistics
|
We study a well known noisy model of the graph isomorphism problem. In this
model, the goal is to perfectly recover the vertex correspondence between two
edge-correlated Erd\H{o}s-R\'{e}nyi random graphs, with an initial seed set of
correctly matched vertex pairs revealed as side information. For seeded
problems, our result provides a significant improvement over previously known
results. We show that it is possible to achieve the information-theoretic limit
of graph sparsity in time polynomial in the number of vertices $n$. Moreover,
we show the number of seeds needed for exact recovery in polynomial-time can be
as low as $n^{3\epsilon}$ in the sparse graph regime (with the average degree
smaller than $n^{\epsilon}$) and $\Omega(\log n)$ in the dense graph regime.
Our results also shed light on the unseeded problem. In particular, we give
sub-exponential time algorithms for sparse models and an $n^{O(\log n)}$
algorithm for dense models for some parameters, including some that are not
covered by recent results of Barak et al.
|
Premise selection with neural networks and distributed representation of
features
|
We present the problem of selecting relevant premises for a proof of a given
statement. When stated as a binary classification task for pairs (conjecture,
axiom), it can be efficiently solved using artificial neural networks. The key
difference between our advance to solve this problem and previous approaches is
the use of just functional signatures of premises. To further improve the
performance of the model, we use dimensionality reduction technique, to replace
long and sparse signature vectors with their compact and dense embedded
versions. These are obtained by firstly defining the concept of a context for
each functor symbol, and then training a simple neural network to predict the
distribution of other functor symbols in the context of this functor. After
training the network, the output of its hidden layer is used to construct a
lower dimensional embedding of a functional signature (for each premise) with a
distributed representation of features. This allows us to use 512-dimensional
embeddings for conjecture-axiom pairs, containing enough information about the
original statements to reach the accuracy of 76.45% in premise selection task,
only with simple two-layer densely connected neural networks.
|
Evaluating and Understanding the Robustness of Adversarial Logit Pairing
|
We evaluate the robustness of Adversarial Logit Pairing, a recently proposed
defense against adversarial examples. We find that a network trained with
Adversarial Logit Pairing achieves 0.6% accuracy in the threat model in which
the defense is considered. We provide a brief overview of the defense and the
threat models/claims considered, as well as a discussion of the methodology and
results of our attack, which may offer insights into the reasons underlying the
vulnerability of ALP to adversarial attack.
|
Structured Point Cloud Data Analysis via Regularized Tensor Regression
for Process Modeling and Optimization
|
Advanced 3D metrology technologies such as Coordinate Measuring Machine (CMM)
and laser 3D scanners have facilitated the collection of massive point cloud
data, beneficial for process monitoring, control and optimization. However, due
to their high dimensionality and structure complexity, modeling and analysis of
point clouds are still a challenge. In this paper, we utilize multilinear
algebra techniques and propose a set of tensor regression approaches to model
the variational patterns of point clouds and to link them to process variables.
The performance of the proposed methods is evaluated through simulations and a
real case study of turning process optimization.
|
Discovering physical concepts with neural networks
|
Despite the success of neural networks at solving concrete physics problems,
their use as a general-purpose tool for scientific discovery is still in its
infancy. Here, we approach this problem by modelling a neural network
architecture after the human physical reasoning process, which has similarities
to representation learning. This allows us to make progress towards the
long-term goal of machine-assisted scientific discovery from experimental data
without making prior assumptions about the system. We apply this method to toy
examples and show that the network finds the physically relevant parameters,
exploits conservation laws to make predictions, and can help to gain conceptual
insights, e.g. Copernicus' conclusion that the solar system is heliocentric.
|
Selective Clustering Annotated using Modes of Projections
|
Selective clustering annotated using modes of projections (SCAMP) is a new
clustering algorithm for data in $\mathbb{R}^p$. SCAMP is motivated from the
point of view of non-parametric mixture modeling. Rather than maximizing a
classification likelihood to determine cluster assignments, SCAMP casts
clustering as a search and selection problem. One consequence of this problem
formulation is that the number of clusters is $\textbf{not}$ a SCAMP tuning
parameter. The search phase of SCAMP consists of finding sub-collections of the
data matrix, called candidate clusters, that obey shape constraints along each
coordinate projection. An extension of the dip test of Hartigan and Hartigan
(1985) is developed to assist the search. Selection occurs by scoring each
candidate cluster with a preference function that quantifies prior belief about
the mixture composition. Clustering proceeds by selecting candidates to
maximize their total preference score. SCAMP concludes by annotating each
selected cluster with labels that describe how cluster-level statistics compare
to certain dataset-level quantities. SCAMP can be run multiple times on a
single data matrix. Comparison of annotations obtained across iterations
provides a measure of clustering uncertainty. Simulation studies and
applications to real data are considered. A C++ implementation with R interface
is $\href{https://github.com/RGLab/scamp}{available\ online}$.
|
A general metric for identifying adversarial images
|
It is well known that a determined adversary can fool a neural network by
making imperceptible adversarial perturbations to an image. Recent studies have
shown that these perturbations can be detected even without information about
the neural network if the strategy taken by the adversary is known beforehand.
Unfortunately, these studies suffer from the generalization limitation -- the
detection method has to be recalibrated every time the adversary changes his
strategy. In this study, we attempt to overcome the generalization limitation
by deriving a metric which reliably identifies adversarial images even when the
approach taken by the adversary is unknown. Our metric leverages key
differences between the spectra of clean and adversarial images when an image
is treated as a matrix. Our metric is able to detect adversarial images across
different datasets and attack strategies without any additional re-calibration.
In addition, our approach provides geometric insights into several unanswered
questions about adversarial perturbations.
|
Message-passing neural networks for high-throughput polymer screening
|
Machine learning methods have shown promise in predicting molecular
properties, and given sufficient training data machine learning approaches can
enable rapid high-throughput virtual screening of large libraries of compounds.
Graph-based neural network architectures have emerged in recent years as the
most successful approach for predictions based on molecular structure, and have
consistently achieved the best performance on benchmark quantum chemical
datasets. However, these models have typically required optimized 3D structural
information for the molecule to achieve the highest accuracy. These 3D
geometries are costly to compute for high levels of theory, limiting the
applicability and practicality of machine learning methods in high-throughput
screening applications. In this study, we present a new database of candidate
molecules for organic photovoltaic applications, comprising approximately
91,000 unique chemical structures.Compared to existing datasets, this dataset
contains substantially larger molecules (up to 200 atoms) as well as
extrapolated properties for long polymer chains. We show that message-passing
neural networks trained with and without 3D structural information for these
molecules achieve similar accuracy, comparable to state-of-the-art methods on
existing benchmark datasets. These results therefore emphasize that for larger
molecules with practical applications, near-optimal prediction results can be
obtained without using optimized 3D geometry as an input. We further show that
learned molecular representations can be leveraged to reduce the training data
required to transfer predictions to a new DFT functional.
|
Applications of Common Entropy for Causal Inference
|
We study the problem of discovering the simplest latent variable that can
make two observed discrete variables conditionally independent. The minimum
entropy required for such a latent is known as common entropy in information
theory. We extend this notion to Renyi common entropy by minimizing the Renyi
entropy of the latent variable. To efficiently compute common entropy, we
propose an iterative algorithm that can be used to discover the trade-off
between the entropy of the latent variable and the conditional mutual
information of the observed variables. We show two applications of common
entropy in causal inference: First, under the assumption that there are no
low-entropy mediators, it can be used to distinguish causation from spurious
correlation among almost all joint distributions on simple causal graphs with
two observed variables. Second, common entropy can be used to improve
constraint-based methods such as PC or FCI algorithms in the small-sample
regime, where these methods are known to struggle. We propose a modification to
these constraint-based methods to assess if a separating set found by these
algorithms is valid using common entropy. We finally evaluate our algorithms on
synthetic and real data to establish their performance.
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.