title
stringlengths 7
246
| abstract
stringlengths 6
3.31k
|
|---|---|
Troubling Trends in Machine Learning Scholarship
|
Collectively, machine learning (ML) researchers are engaged in the creation
and dissemination of knowledge about data-driven algorithms. In a given paper,
researchers might aspire to any subset of the following goals, among others: to
theoretically characterize what is learnable, to obtain understanding through
empirically rigorous experiments, or to build a working system that has high
predictive accuracy. While determining which knowledge warrants inquiry may be
subjective, once the topic is fixed, papers are most valuable to the community
when they act in service of the reader, creating foundational knowledge and
communicating as clearly as possible.
Recent progress in machine learning comes despite frequent departures from
these ideals. In this paper, we focus on the following four patterns that
appear to us to be trending in ML scholarship: (i) failure to distinguish
between explanation and speculation; (ii) failure to identify the sources of
empirical gains, e.g., emphasizing unnecessary modifications to neural
architectures when gains actually stem from hyper-parameter tuning; (iii)
mathiness: the use of mathematics that obfuscates or impresses rather than
clarifies, e.g., by confusing technical and non-technical concepts; and (iv)
misuse of language, e.g., by choosing terms of art with colloquial connotations
or by overloading established technical terms.
While the causes behind these patterns are uncertain, possibilities include
the rapid expansion of the community, the consequent thinness of the reviewer
pool, and the often-misaligned incentives between scholarship and short-term
measures of success (e.g., bibliometrics, attention, and entrepreneurial
opportunity). While each pattern offers a corresponding remedy (don't do it),
we also discuss some speculative suggestions for how the community might combat
these trends.
|
Weakly-Supervised Convolutional Neural Networks for Multimodal Image
Registration
|
One of the fundamental challenges in supervised learning for multimodal image
registration is the lack of ground-truth for voxel-level spatial
correspondence. This work describes a method to infer voxel-level
transformation from higher-level correspondence information contained in
anatomical labels. We argue that such labels are more reliable and practical to
obtain for reference sets of image pairs than voxel-level correspondence.
Typical anatomical labels of interest may include solid organs, vessels, ducts,
structure boundaries and other subject-specific ad hoc landmarks. The proposed
end-to-end convolutional neural network approach aims to predict displacement
fields to align multiple labelled corresponding structures for individual image
pairs during the training, while only unlabelled image pairs are used as the
network input for inference. We highlight the versatility of the proposed
strategy, for training, utilising diverse types of anatomical labels, which
need not to be identifiable over all training image pairs. At inference, the
resulting 3D deformable image registration algorithm runs in real-time and is
fully-automated without requiring any anatomical labels or initialisation.
Several network architecture variants are compared for registering T2-weighted
magnetic resonance images and 3D transrectal ultrasound images from prostate
cancer patients. A median target registration error of 3.6 mm on landmark
centroids and a median Dice of 0.87 on prostate glands are achieved from
cross-validation experiments, in which 108 pairs of multimodal images from 76
patients were tested with high-quality anatomical labels.
|
Talk the Walk: Navigating New York City through Grounded Dialogue
|
We introduce "Talk The Walk", the first large-scale dialogue dataset grounded
in action and perception. The task involves two agents (a "guide" and a
"tourist") that communicate via natural language in order to achieve a common
goal: having the tourist navigate to a given target location. The task and
dataset, which are described in detail, are challenging and their full solution
is an open problem that we pose to the community. We (i) focus on the task of
tourist localization and develop the novel Masked Attention for Spatial
Convolutions (MASC) mechanism that allows for grounding tourist utterances into
the guide's map, (ii) show it yields significant improvements for both emergent
and natural language communication, and (iii) using this method, we establish
non-trivial baselines on the full task.
|
Tractable $n$-Metrics for Multiple Graphs
|
Graphs are used in almost every scientific discipline to express relations
among a set of objects. Algorithms that compare graphs, and output a closeness
score, or a correspondence among their nodes, are thus extremely important.
Despite the large amount of work done, many of the scalable algorithms to
compare graphs do not produce closeness scores that satisfy the intuitive
properties of metrics. This is problematic since non-metrics are known to
degrade the performance of algorithms such as distance-based clustering of
graphs (Stratis and Bento 2018). On the other hand, the use of metrics
increases the performance of several machine learning tasks (Indyk et al. 1999,
Clarkson et al. 1999, Angiulli et al. 2002, Ackermann et al. 2010). In this
paper, we introduce a new family of multi-distances (a distance between more
than two elements) that satisfies a generalization of the properties of metrics
to multiple elements. In the context of comparing graphs, we are the first to
show the existence of multi-distances that simultaneously incorporate the
useful property of alignment consistency (Nguyen et al. 2011), and a
generalized metric property. Furthermore, we show that these multi-distances
can be relaxed to convex optimization problems, without losing the generalized
metric property.
|
Ensemble Kalman Filtering for Online Gaussian Process Regression and
Learning
|
Gaussian process regression is a machine learning approach which has been
shown its power for estimation of unknown functions. However, Gaussian
processes suffer from high computational complexity, as in a basic form they
scale cubically with the number of observations. Several approaches based on
inducing points were proposed to handle this problem in a static context. These
methods though face challenges with real-time tasks and when the data is
received sequentially over time. In this paper, a novel online algorithm for
training sparse Gaussian process models is presented. It treats the mean and
hyperparameters of the Gaussian process as the state and parameters of the
ensemble Kalman filter, respectively. The online evaluation of the parameters
and the state is performed on new upcoming samples of data. This procedure
iteratively improves the accuracy of parameter estimates. The ensemble Kalman
filter reduces the computational complexity required to obtain predictions with
Gaussian processes preserving the accuracy level of these predictions. The
performance of the proposed method is demonstrated on the synthetic dataset and
real large dataset of UK house prices.
|
Online Scoring with Delayed Information: A Convex Optimization Viewpoint
|
We consider a system where agents enter in an online fashion and are
evaluated based on their attributes or context vectors. There can be practical
situations where this context is partially observed, and the unobserved part
comes after some delay. We assume that an agent, once left, cannot re-enter the
system. Therefore, the job of the system is to provide an estimated score for
the agent based on her instantaneous score and possibly some inference of the
instantaneous score over the delayed score. In this paper, we estimate the
delayed context via an online convex game between the agent and the system. We
argue that the error in the score estimate accumulated over $T$ iterations is
small if the regret of the online convex game is small. Further, we leverage
side information about the delayed context in the form of a correlation
function with the known context. We consider the settings where the delay is
fixed or arbitrarily chosen by an adversary. Furthermore, we extend the
formulation to the setting where the contexts are drawn from some Banach space.
Overall, we show that the average penalty for not knowing the delayed context
while making a decision scales with $\mathcal{O}(\frac{1}{\sqrt{T}})$, where
this can be improved to $\mathcal{O}(\frac{\log T}{T})$ under special setting.
|
Process Monitoring Using Maximum Sequence Divergence
|
Process Monitoring involves tracking a system's behaviors, evaluating the
current state of the system, and discovering interesting events that require
immediate actions. In this paper, we consider monitoring temporal system state
sequences to help detect the changes of dynamic systems, check the divergence
of the system development, and evaluate the significance of the deviation. We
begin with discussions of data reduction, symbolic data representation, and the
anomaly detection in temporal discrete sequences. Time-series representation
methods are also discussed and used in this paper to discretize raw data into
sequences of system states. Markov Chains and stationary state distributions
are continuously generated from temporal sequences to represent snapshots of
the system dynamics in different time frames. We use generalized Jensen-Shannon
Divergence as the measure to monitor changes of the stationary symbol
probability distributions and evaluate the significance of system deviations.
We prove that the proposed approach is able to detect deviations of the systems
we monitor and assess the deviation significance in probabilistic manner.
|
Towards Non-Parametric Learning to Rank
|
This paper studies a stylized, yet natural, learning-to-rank problem and
points out the critical incorrectness of a widely used nearest neighbor
algorithm. We consider a model with $n$ agents (users) $\{x_i\}_{i \in [n]}$
and $m$ alternatives (items) $\{y_j\}_{j \in [m]}$, each of which is associated
with a latent feature vector. Agents rank items nondeterministically according
to the Plackett-Luce model, where the higher the utility of an item to the
agent, the more likely this item will be ranked high by the agent. Our goal is
to find neighbors of an arbitrary agent or alternative in the latent space.
We first show that the Kendall-tau distance based kNN produces incorrect
results in our model. Next, we fix the problem by introducing a new algorithm
with features constructed from "global information" of the data matrix. Our
approach is in sharp contrast to most existing feature engineering methods.
Finally, we design another new algorithm identifying similar alternatives. The
construction of alternative features can be done using "local information,"
highlighting the algorithmic difference between finding similar agents and
similar alternatives.
|
On Training Recurrent Networks with Truncated Backpropagation Through
Time in Speech Recognition
|
Recurrent neural networks have been the dominant models for many speech and
language processing tasks. However, we understand little about the behavior and
the class of functions recurrent networks can realize. Moreover, the heuristics
used during training complicate the analyses. In this paper, we study recurrent
networks' ability to learn long-term dependency in the context of speech
recognition. We consider two decoding approaches, online and batch decoding,
and show the classes of functions to which the decoding approaches correspond.
We then draw a connection between batch decoding and a popular training
approach for recurrent networks, truncated backpropagation through time.
Changing the decoding approach restricts the amount of past history recurrent
networks can use for prediction, allowing us to analyze their ability to
remember. Empirically, we utilize long-term dependency in subphonetic states,
phonemes, and words, and show how the design decisions, such as the decoding
approach, lookahead, context frames, and consecutive prediction, characterize
the behavior of recurrent networks. Finally, we draw a connection between
Markov processes and vanishing gradients. These results have implications for
studying the long-term dependency in speech data and how these properties are
learned by recurrent networks.
|
IGLOO: Slicing the Features Space to Represent Sequences
|
Historically, Recurrent neural networks (RNNs) and its variants such as LSTM
and GRU and more recently Transformers have been the standard go-to components
when processing sequential data with neural networks. One notable issue is the
relative difficulty to deal with long sequences (i.e. more than 20,000 steps).
We introduce IGLOO, a new neural network architecture which aims at being
efficient for short sequences but also at being able to deal with long
sequences. IGLOOs core idea is to use the relationships between non-local
patches sliced out of the features maps of successively applied convolutions to
build a representation for the sequence. We show that the model can deal with
dependencies of more than 20,000 steps in a reasonable time frame. We stress
test IGLOO on the copy-memory and addition tasks, as well as permuted MNIST
(98.4%). For a larger task we apply this new structure to the Wikitext-2
dataset Merity et al. (2017b) and achieve a perplexity in line with baseline
Transformers but lower than baseline AWD-LSTM. We also present how IGLOO is
already used today in production for bioinformatics tasks.
|
Hierarchical Visualization of Materials Space with Graph Convolutional
Neural Networks
|
The combination of high throughput computation and machine learning has led
to a new paradigm in materials design by allowing for the direct screening of
vast portions of structural, chemical, and property space. The use of these
powerful techniques leads to the generation of enormous amounts of data, which
in turn calls for new techniques to efficiently explore and visualize the
materials space to help identify underlying patterns. In this work, we develop
a unified framework to hierarchically visualize the compositional and
structural similarities between materials in an arbitrary material space with
representations learned from different layers of graph convolutional neural
networks. We demonstrate the potential for such a visualization approach by
showing that patterns emerge automatically that reflect similarities at
different scales in three representative classes of materials: perovskites,
elemental boron, and general inorganic crystals, covering material spaces of
different compositions, structures, and both. For perovskites, elemental
similarities are learned that reflects multiple aspects of atom properties. For
elemental boron, structural motifs emerge automatically showing characteristic
boron local environments. For inorganic crystals, the similarity and stability
of local coordination environments are shown combining different center and
neighbor atoms. The method could help transition to a data-centered exploration
of materials space in automated materials design.
|
Interpreting and Explaining Deep Neural Networks for Classification of
Audio Signals
|
Interpretability of deep neural networks is a recently emerging area of
machine learning research targeting a better understanding of how models
perform feature selection and derive their classification decisions. This paper
explores the interpretability of neural networks in the audio domain by using
the previously proposed technique of layer-wise relevance propagation (LRP). We
present a novel audio dataset of English spoken digits which we use for
classification tasks on spoken digits and speaker's gender. We use LRP to
identify relevant features for two neural network architectures that process
either waveform or spectrogram representations of the data. Based on the
relevance scores obtained from LRP, hypotheses about the neural networks'
feature selection are derived and subsequently tested through systematic
manipulations of the input data. The results confirm that the networks are
highly reliant on features marked as relevant by LRP.
|
A GPU-Oriented Algorithm Design for Secant-Based Dimensionality
Reduction
|
Dimensionality-reduction techniques are a fundamental tool for extracting
useful information from high-dimensional data sets. Because secant sets encode
manifold geometry, they are a useful tool for designing meaningful
data-reduction algorithms. In one such approach, the goal is to construct a
projection that maximally avoids secant directions and hence ensures that
distinct data points are not mapped too close together in the reduced space.
This type of algorithm is based on a mathematical framework inspired by the
constructive proof of Whitney's embedding theorem from differential topology.
Computing all (unit) secants for a set of points is by nature computationally
expensive, thus opening the door for exploitation of GPU architecture for
achieving fast versions of these algorithms. We present a polynomial-time
data-reduction algorithm that produces a meaningful low-dimensional
representation of a data set by iteratively constructing improved projections
within the framework described above. Key to our algorithm design and
implementation is the use of GPUs which, among other things, minimizes the
computational time required for the calculation of all secant lines. One goal
of this report is to share ideas with GPU experts and to discuss a class of
mathematical algorithms that may be of interest to the broader GPU community.
|
A New Variational Model for Binary Classification in the Supervised
Learning Context
|
We examine the supervised learning problem in its continuous setting and give
a general optimality condition through techniques of functional analysis and
the calculus of variations. This enables us to solve the optimality condition
for the desired function u numerically and make several comparisons with other
widely utilized supervised learning models. We employ the accuracy and area
under the receiver operating characteristic curve as metrics of the
performance. Finally, 3 analyses are conducted based on these two mentioned
metrics where we compare the models and make conclusions to determine whether
or not our method is competitive.
|
Developing Brain Atlas through Deep Learning
|
Neuroscientists have devoted significant effort into the creation of standard
brain reference atlases for high-throughput registration of anatomical regions
of interest. However, variability in brain size and form across individuals
poses a significant challenge for such reference atlases. To overcome these
limitations, we introduce a fully automated deep neural network-based method
(SeBRe) for registration through Segmenting Brain Regions of interest with
minimal human supervision. We demonstrate the validity of our method on brain
images from different mouse developmental time points, across a range of
neuronal markers and imaging modalities. We further assess the performance of
our method on images from MR-scanned human brains. Our registration method can
accelerate brain-wide exploration of region-specific changes in brain
development and, by simply segmenting brain regions of interest for
high-throughput brain-wide analysis, provides an alternative to existing
complex brain registration techniques.
|
Predicting property damage from tornadoes with zero-inflated neural
networks
|
Tornadoes are the most violent of all atmospheric storms. In a typical year,
the United States experiences hundreds of tornadoes with associated damages on
the order of one billion dollars. Community preparation and resilience would
benefit from accurate predictions of these economic losses, particularly as
populations in tornado-prone areas increase in density and extent. Here, we use
a zero-inflated modeling approach and artificial neural networks to predict
tornado-induced property damage using publicly available data. We developed a
neural network that predicts whether a tornado will cause property damage
(out-of-sample accuracy = 0.821 and area under the receiver operating
characteristic curve, AUROC, = 0.872). Conditional on a tornado causing damage,
another neural network predicts the amount of damage (out-of-sample mean
squared error = 0.0918 and R2 = 0.432). When used together, these two models
function as a zero-inflated log-normal regression with hidden layers. From the
best-performing models, we provide static and interactive gridded maps of
monthly predicted probabilities of damage and property damages for the year
2019. Two primary weaknesses include (1) model fitting requires log-scale data
which leads to large natural-scale residuals and (2) beginning tornado
coordinates were utilized rather than tornado paths. Ultimately, this is the
first known study to directly model tornado-induced property damages, and all
data, code, and tools are publicly available. The predictive capacity of this
model along with an interactive interface may provide an opportunity for
science-informed tornado disaster planning.
|
SceneEDNet: A Deep Learning Approach for Scene Flow Estimation
|
Estimating scene flow in RGB-D videos is attracting much interest of the
computer vision researchers, due to its potential applications in robotics. The
state-of-the-art techniques for scene flow estimation, typically rely on the
knowledge of scene structure of the frame and the correspondence between
frames. However, with the increasing amount of RGB-D data captured from
sophisticated sensors like Microsoft Kinect, and the recent advances in the
area of sophisticated deep learning techniques, introduction of an efficient
deep learning technique for scene flow estimation, is becoming important. This
paper introduces a first effort to apply a deep learning method for direct
estimation of scene flow by presenting a fully convolutional neural network
with an encoder-decoder (ED) architecture. The proposed network SceneEDNet
involves estimation of three dimensional motion vectors of all the scene points
from sequence of stereo images. The training for direct estimation of scene
flow is done using consecutive pairs of stereo images and corresponding scene
flow ground truth. The proposed architecture is applied on a huge dataset and
provides meaningful results.
|
An Adaptive Learning Method of Restricted Boltzmann Machine by Neuron
Generation and Annihilation Algorithm
|
Restricted Boltzmann Machine (RBM) is a generative stochastic energy-based
model of artificial neural network for unsupervised learning. Recently, RBM is
well known to be a pre-training method of Deep Learning. In addition to visible
and hidden neurons, the structure of RBM has a number of parameters such as the
weights between neurons and the coefficients for them. Therefore, we may meet
some difficulties to determine an optimal network structure to analyze big
data. In order to evade the problem, we investigated the variance of parameters
to find an optimal structure during learning. For the reason, we should check
the variance of parameters to cause the fluctuation for energy function in RBM
model. In this paper, we propose the adaptive learning method of RBM that can
discover an optimal number of hidden neurons according to the training
situation by applying the neuron generation and annihilation algorithm. In this
method, a new hidden neuron is generated if the energy function is not still
converged and the variance of the parameters is large. Moreover, the
inactivated hidden neuron will be annihilated if the neuron does not affect the
learning situation. The experimental results for some benchmark data sets were
discussed in this paper.
|
Neural Task Graphs: Generalizing to Unseen Tasks from a Single Video
Demonstration
|
Our goal is to generate a policy to complete an unseen task given just a
single video demonstration of the task in a given domain. We hypothesize that
to successfully generalize to unseen complex tasks from a single video
demonstration, it is necessary to explicitly incorporate the compositional
structure of the tasks into the model. To this end, we propose Neural Task
Graph (NTG) Networks, which use conjugate task graph as the intermediate
representation to modularize both the video demonstration and the derived
policy. We empirically show NTG achieves inter-task generalization on two
complex tasks: Block Stacking in BulletPhysics and Object Collection in
AI2-THOR. NTG improves data efficiency with visual input as well as achieve
strong generalization without the need for dense hierarchical supervision. We
further show that similar performance trends hold when applied to real-world
data. We show that NTG can effectively predict task structure on the JIGSAWS
surgical dataset and generalize to unseen tasks.
|
Easing Embedding Learning by Comprehensive Transcription of
Heterogeneous Information Networks
|
Heterogeneous information networks (HINs) are ubiquitous in real-world
applications. In the meantime, network embedding has emerged as a convenient
tool to mine and learn from networked data. As a result, it is of interest to
develop HIN embedding methods. However, the heterogeneity in HINs introduces
not only rich information but also potentially incompatible semantics, which
poses special challenges to embedding learning in HINs. With the intention to
preserve the rich yet potentially incompatible information in HIN embedding, we
propose to study the problem of comprehensive transcription of heterogeneous
information networks. The comprehensive transcription of HINs also provides an
easy-to-use approach to unleash the power of HINs, since it requires no
additional supervision, expertise, or feature engineering. To cope with the
challenges in the comprehensive transcription of HINs, we propose the HEER
algorithm, which embeds HINs via edge representations that are further coupled
with properly-learned heterogeneous metrics. To corroborate the efficacy of
HEER, we conducted experiments on two large-scale real-words datasets with an
edge reconstruction task and multiple case studies. Experiment results
demonstrate the effectiveness of the proposed HEER model and the utility of
edge representations and heterogeneous metrics. The code and data are available
at https://github.com/GentleZhu/HEER.
|
Multiresolution Tree Networks for 3D Point Cloud Processing
|
We present multiresolution tree-structured networks to process point clouds
for 3D shape understanding and generation tasks. Our network represents a 3D
shape as a set of locality-preserving 1D ordered list of points at multiple
resolutions. This allows efficient feed-forward processing through 1D
convolutions, coarse-to-fine analysis through a multi-grid architecture, and it
leads to faster convergence and small memory footprint during training. The
proposed tree-structured encoders can be used to classify shapes and outperform
existing point-based architectures on shape classification benchmarks, while
tree-structured decoders can be used for generating point clouds directly and
they outperform existing approaches for image-to-shape inference tasks learned
using the ShapeNet dataset. Our model also allows unsupervised learning of
point-cloud based shapes by using a variational autoencoder, leading to
higher-quality generated shapes.
|
Privacy and Fairness in Recommender Systems via Adversarial Training of
User Representations
|
Latent factor models for recommender systems represent users and items as low
dimensional vectors. Privacy risks of such systems have previously been studied
mostly in the context of recovery of personal information in the form of usage
records from the training data. However, the user representations themselves
may be used together with external data to recover private user information
such as gender and age. In this paper we show that user vectors calculated by a
common recommender system can be exploited in this way. We propose the
privacy-adversarial framework to eliminate such leakage of private information,
and study the trade-off between recommender performance and leakage both
theoretically and empirically using a benchmark dataset. An advantage of the
proposed method is that it also helps guarantee fairness of results, since all
implicit knowledge of a set of attributes is scrubbed from the representations
used by the model, and thus can't enter into the decision making. We discuss
further applications of this method towards the generation of deeper and more
insightful recommendations.
|
DLOPT: Deep Learning Optimization Library
|
Deep learning hyper-parameter optimization is a tough task. Finding an
appropriate network configuration is a key to success, however most of the
times this labor is roughly done. In this work we introduce a novel library to
tackle this problem, the Deep Learning Optimization Library: DLOPT. We briefly
describe its architecture and present a set of use examples. This is an open
source project developed under the GNU GPL v3 license and it is freely
available at https://github.com/acamero/dlopt
|
Algebraic Equivalence of Linear Structural Equation Models
|
Despite their popularity, many questions about the algebraic constraints
imposed by linear structural equation models remain open problems. For causal
discovery, two of these problems are especially important: the enumeration of
the constraints imposed by a model, and deciding whether two graphs define the
same statistical model. We show how the half-trek criterion can be used to make
progress in both of these problems. We apply our theoretical results to a
small-scale model selection problem, and find that taking the additional
algebraic constraints into account may lead to significant improvements in
model selection accuracy.
|
Dual optimization for convex constrained objectives without the
gradient-Lipschitz assumption
|
The minimization of convex objectives coming from linear supervised learning
problems, such as penalized generalized linear models, can be formulated as
finite sums of convex functions. For such problems, a large set of stochastic
first-order solvers based on the idea of variance reduction are available and
combine both computational efficiency and sound theoretical guarantees (linear
convergence rates). Such rates are obtained under both gradient-Lipschitz and
strong convexity assumptions. Motivated by learning problems that do not meet
the gradient-Lipschitz assumption, such as linear Poisson regression, we work
under another smoothness assumption, and obtain a linear convergence rate for a
shifted version of Stochastic Dual Coordinate Ascent (SDCA) that improves the
current state-of-the-art. Our motivation for considering a solver working on
the Fenchel-dual problem comes from the fact that such objectives include many
linear constraints, that are easier to deal with in the dual. Our approach and
theoretical findings are validated on several datasets, for Poisson regression
and another objective coming from the negative log-likelihood of the Hawkes
process, which is a family of models which proves extremely useful for the
modeling of information propagation in social networks and causality inference.
|
An Empirical Approach For Probing the Definiteness of Kernels
|
Models like support vector machines or Gaussian process regression often
require positive semi-definite kernels. These kernels may be based on distance
functions. While definiteness is proven for common distances and kernels, a
proof for a new kernel may require too much time and effort for users who
simply aim at practical usage. Furthermore, designing definite distances or
kernels may be equally intricate. Finally, models can be enabled to use
indefinite kernels. This may deteriorate the accuracy or computational cost of
the model. Hence, an efficient method to determine definiteness is required. We
propose an empirical approach. We show that sampling as well as optimization
with an evolutionary algorithm may be employed to determine definiteness. We
provide a proof-of-concept with 16 different distance measures for
permutations. Our approach allows to disprove definiteness if a respective
counter-example is found. It can also provide an estimate of how likely it is
to obtain indefinite kernel matrices. This provides a simple, efficient tool to
decide whether additional effort should be spent on designing/selecting a more
suitable kernel or algorithm.
|
Bandits with Side Observations: Bounded vs. Logarithmic Regret
|
We consider the classical stochastic multi-armed bandit but where, from time
to time and roughly with frequency $\epsilon$, an extra observation is gathered
by the agent for free. We prove that, no matter how small $\epsilon$ is the
agent can ensure a regret uniformly bounded in time.
More precisely, we construct an algorithm with a regret smaller than $\sum_i
\frac{\log(1/\epsilon)}{\Delta_i}$, up to multiplicative constant and loglog
terms. We also prove a matching lower-bound, stating that no reasonable
algorithm can outperform this quantity.
|
Small-Variance Asymptotics for Nonparametric Bayesian Overlapping
Stochastic Blockmodels
|
The latent feature relational model (LFRM) is a generative model for
graph-structured data to learn a binary vector representation for each node in
the graph. The binary vector denotes the node's membership in one or more
communities. At its core, the LFRM miller2009nonparametric is an overlapping
stochastic blockmodel, which defines the link probability between any pair of
nodes as a bilinear function of their community membership vectors. Moreover,
using a nonparametric Bayesian prior (Indian Buffet Process) enables learning
the number of communities automatically from the data. However, despite its
appealing properties, inference in LFRM remains a challenge and is typically
done via MCMC methods. This can be slow and may take a long time to converge.
In this work, we develop a small-variance asymptotics based framework for the
non-parametric Bayesian LFRM. This leads to an objective function that retains
the nonparametric Bayesian flavor of LFRM, while enabling us to design
deterministic inference algorithms for this model, that are easy to implement
(using generic or specialized optimization routines) and are fast in practice.
Our results on several benchmark datasets demonstrate that our algorithm is
competitive to methods such as MCMC, while being much faster.
|
A Game-Based Approximate Verification of Deep Neural Networks with
Provable Guarantees
|
Despite the improved accuracy of deep neural networks, the discovery of
adversarial examples has raised serious safety concerns. In this paper, we
study two variants of pointwise robustness, the maximum safe radius problem,
which for a given input sample computes the minimum distance to an adversarial
example, and the feature robustness problem, which aims to quantify the
robustness of individual features to adversarial perturbations. We demonstrate
that, under the assumption of Lipschitz continuity, both problems can be
approximated using finite optimisation by discretising the input space, and the
approximation has provable guarantees, i.e., the error is bounded. We then show
that the resulting optimisation problems can be reduced to the solution of
two-player turn-based games, where the first player selects features and the
second perturbs the image within the feature. While the second player aims to
minimise the distance to an adversarial example, depending on the optimisation
objective the first player can be cooperative or competitive. We employ an
anytime approach to solve the games, in the sense of approximating the value of
a game by monotonically improving its upper and lower bounds. The Monte Carlo
tree search algorithm is applied to compute upper bounds for both games, and
the Admissible A* and the Alpha-Beta Pruning algorithms are, respectively, used
to compute lower bounds for the maximum safety radius and feature robustness
games. When working on the upper bound of the maximum safe radius problem, our
tool demonstrates competitive performance against existing adversarial example
crafting algorithms. Furthermore, we show how our framework can be deployed to
evaluate pointwise robustness of neural networks in safety-critical
applications such as traffic sign recognition in self-driving cars.
|
Window Opening Model using Deep Learning Methods
|
Occupant behavior (OB) and in particular window openings need to be
considered in building performance simulation (BPS), in order to realistically
model the indoor climate and energy consumption for heating ventilation and air
conditioning (HVAC). However, the proposed OB window opening models are often
biased towards the over-represented class where windows remained closed. In
addition, they require tuning for each occupant which can not be efficiently
scaled to the increased number of occupants. This paper presents a window
opening model for commercial buildings using deep learning methods. The model
is trained using data from occupants from an office building in Germany. In
total the model is evaluated using almost 20 mio. data points from 3
independent buildings, located in Aachen, Frankfurt and Philadelphia.
Eventually, the results of 3100 core hours of model development are summarized,
which makes this study the largest of its kind in window states modeling.
Additionally, the practical potential of the proposed model was tested by
incorporating it in the Modelica-based thermal building simulation. The
resulting evaluation accuracy and F1 scores on the office buildings ranged
between 86-89 % and 0.53-0.65 respectively. The performance dropped around 15 %
points in case of sparse input data, while the F1 score remained high.
|
Foreign English Accent Adjustment by Learning Phonetic Patterns
|
State-of-the-art automatic speech recognition (ASR) systems struggle with the
lack of data for rare accents. For sufficiently large datasets, neural engines
tend to outshine statistical models in most natural language processing
problems. However, a speech accent remains a challenge for both approaches.
Phonologists manually create general rules describing a speaker's accent, but
their results remain underutilized. In this paper, we propose a model that
automatically retrieves phonological generalizations from a small dataset. This
method leverages the difference in pronunciation between a particular dialect
and General American English (GAE) and creates new accented samples of words.
The proposed model is able to learn all generalizations that previously were
manually obtained by phonologists. We use this statistical method to generate a
million phonological variations of words from the CMU Pronouncing Dictionary
and train a sequence-to-sequence RNN to recognize accented words with 59%
accuracy.
|
Handling Incomplete Heterogeneous Data using VAEs
|
Variational autoencoders (VAEs), as well as other generative models, have
been shown to be efficient and accurate for capturing the latent structure of
vast amounts of complex high-dimensional data. However, existing VAEs can still
not directly handle data that are heterogenous (mixed continuous and discrete)
or incomplete (with missing data at random), which is indeed common in
real-world applications. In this paper, we propose a general framework to
design VAEs suitable for fitting incomplete heterogenous data. The proposed
HI-VAE includes likelihood models for real-valued, positive real valued,
interval, categorical, ordinal and count data, and allows accurate estimation
(and potentially imputation) of missing data. Furthermore, HI-VAE presents
competitive predictive performance in supervised tasks, outperforming
supervised models when trained on incomplete data.
|
Deep Learning for Audio Transcription on Low-Resource Datasets
|
In training a deep learning system to perform audio transcription, two
practical problems may arise. Firstly, most datasets are weakly labelled,
having only a list of events present in each recording without any temporal
information for training. Secondly, deep neural networks need a very large
amount of labelled training data to achieve good quality performance, yet in
practice it is difficult to collect enough samples for most classes of
interest. In this paper, we propose factorising the final task of audio
transcription into multiple intermediate tasks in order to improve the training
performance when dealing with this kind of low-resource datasets. We evaluate
three data-efficient approaches of training a stacked convolutional and
recurrent neural network for the intermediate tasks. Our results show that
different methods of training have different advantages and disadvantages.
|
Deterministic Policy Gradients With General State Transitions
|
We study a reinforcement learning setting, where the state transition
function is a convex combination of a stochastic continuous function and a
deterministic function. Such a setting generalizes the widely-studied
stochastic state transition setting, namely the setting of deterministic policy
gradient (DPG).
We firstly give a simple example to illustrate that the deterministic policy
gradient may be infinite under deterministic state transitions, and introduce a
theoretical technique to prove the existence of the policy gradient in this
generalized setting. Using this technique, we prove that the deterministic
policy gradient indeed exists for a certain set of discount factors, and
further prove two conditions that guarantee the existence for all discount
factors. We then derive a closed form of the policy gradient whenever exists.
Furthermore, to overcome the challenge of high sample complexity of DPG in this
setting, we propose the Generalized Deterministic Policy Gradient (GDPG)
algorithm. The main innovation of the algorithm is a new method of applying
model-based techniques to the model-free algorithm, the deep deterministic
policy gradient algorithm (DDPG). GDPG optimize the long-term rewards of the
model-based augmented MDP subject to a constraint that the long-rewards of the
MDP is less than the original one.
We finally conduct extensive experiments comparing GDPG with state-of-the-art
methods and the direct model-based extension method of DDPG on several standard
continuous control benchmarks. Results demonstrate that GDPG substantially
outperforms DDPG, the model-based extension of DDPG and other baselines in
terms of both convergence and long-term rewards in most environments.
|
Recurrent Auto-Encoder Model for Large-Scale Industrial Sensor Signal
Analysis
|
Recurrent auto-encoder model summarises sequential data through an encoder
structure into a fixed-length vector and then reconstructs the original
sequence through the decoder structure. The summarised vector can be used to
represent time series features. In this paper, we propose relaxing the
dimensionality of the decoder output so that it performs partial
reconstruction. The fixed-length vector therefore represents features in the
selected dimensions only. In addition, we propose using rolling fixed window
approach to generate training samples from unbounded time series data. The
change of time series features over time can be summarised as a smooth
trajectory path. The fixed-length vectors are further analysed using additional
visualisation and unsupervised clustering techniques. The proposed method can
be applied in large-scale industrial processes for sensors signal analysis
purpose, where clusters of the vector representations can reflect the operating
states of the industrial system.
|
Geometric Generalization Based Zero-Shot Learning Dataset Infinite
World: Simple Yet Powerful
|
Raven's Progressive Matrices are one of the widely used tests in evaluating
the human test taker's fluid intelligence. Analogously, this paper introduces
geometric generalization based zero-shot learning tests to measure the rapid
learning ability and the internal consistency of deep generative models. Our
empirical research analysis on state-of-the-art generative models discern their
ability to generalize concepts across classes. In the process, we introduce
Infinite World, an evaluable, scalable, multi-modal, light-weight dataset and
Zero-Shot Intelligence Metric ZSI. The proposed tests condenses human-level
spatial and numerical reasoning tasks to its simplistic geometric forms. The
dataset is scalable to a theoretical limit of infinity, in numerical features
of the generated geometric figures, image size and in quantity. We
systematically analyze state-of-the-art model's internal consistency, identify
their bottlenecks and propose a pro-active optimization method for few-shot and
zero-shot learning.
|
Understanding VAEs in Fisher-Shannon Plane
|
In information theory, Fisher information and Shannon information (entropy)
are respectively used to quantify the uncertainty associated with the
distribution modeling and the uncertainty in specifying the outcome of given
variables. These two quantities are complementary and are jointly applied to
information behavior analysis in most cases. The uncertainty property in
information asserts a fundamental trade-off between Fisher information and
Shannon information, which enlightens us the relationship between the encoder
and the decoder in variational auto-encoders (VAEs). In this paper, we
investigate VAEs in the Fisher-Shannon plane and demonstrate that the
representation learning and the log-likelihood estimation are intrinsically
related to these two information quantities. Through extensive qualitative and
quantitative experiments, we provide with a better comprehension of VAEs in
tasks such as high-resolution reconstruction, and representation learning in
the perspective of Fisher information and Shannon information. We further
propose a variant of VAEs, termed as Fisher auto-encoder (FAE), for practical
needs to balance Fisher information and Shannon information. Our experimental
results have demonstrated its promise in improving the reconstruction accuracy
and avoiding the non-informative latent code as occurred in previous works.
|
Network Classification in Temporal Networks Using Motifs
|
Network classification has a variety of applications, such as detecting
communities within networks and finding similarities between those representing
different aspects of the real world. However, most existing work in this area
focus on examining static undirected networks without considering directed
edges or temporality. In this paper, we propose a new methodology that utilizes
feature representation for network classification based on the temporal motif
distribution of the network and a null model for comparing against random
graphs. Experimental results show that our method improves accuracy by up
$10\%$ compared to the state-of-the-art embedding method in network
classification, for tasks such as classifying network type, identifying
communities in email exchange network, and identifying users given their
app-switching behaviors.
|
Scalable Sparse Subspace Clustering via Ordered Weighted $\ell_1$
Regression
|
The main contribution of the paper is a new approach to subspace clustering
that is significantly more computationally efficient and scalable than existing
state-of-the-art methods. The central idea is to modify the regression
technique in sparse subspace clustering (SSC) by replacing the $\ell_1$
minimization with a generalization called Ordered Weighted $\ell_1$ (OWL)
minimization which performs simultaneous regression and clustering of
correlated variables. Using random geometric graph theory, we prove that OWL
regression selects more points within each subspace, resulting in better
clustering results. This allows for accurate subspace clustering based on
regression solutions for only a small subset of the total dataset,
significantly reducing the computational complexity compared to SSC. In
experiments, we find that our OWL approach can achieve a speedup of 20$\times$
to 30$\times$ for synthetic problems and 4$\times$ to 8$\times$ on real data
problems.
|
Representation Learning with Contrastive Predictive Coding
|
While supervised learning has enabled great progress in many applications,
unsupervised learning has not seen such widespread adoption, and remains an
important and challenging endeavor for artificial intelligence. In this work,
we propose a universal unsupervised learning approach to extract useful
representations from high-dimensional data, which we call Contrastive
Predictive Coding. The key insight of our model is to learn such
representations by predicting the future in latent space by using powerful
autoregressive models. We use a probabilistic contrastive loss which induces
the latent space to capture information that is maximally useful to predict
future samples. It also makes the model tractable by using negative sampling.
While most prior work has focused on evaluating representations for a
particular modality, we demonstrate that our approach is able to learn useful
representations achieving strong performance on four distinct domains: speech,
images, text and reinforcement learning in 3D environments.
|
Navigating Diverse Data Science Learning: Critical Reflections Towards
Future Practice
|
Data Science is currently a popular field of science attracting expertise
from very diverse backgrounds. Current learning practices need to acknowledge
this and adapt to it. This paper summarises some experiences relating to such
learning approaches from teaching a postgraduate Data Science module, and draws
some learned lessons that are of relevance to others teaching Data Science.
|
Latent Alignment and Variational Attention
|
Neural attention has become central to many state-of-the-art models in
natural language processing and related domains. Attention networks are an
easy-to-train and effective method for softly simulating alignment; however,
the approach does not marginalize over latent alignments in a probabilistic
sense. This property makes it difficult to compare attention to other alignment
approaches, to compose it with probabilistic models, and to perform posterior
inference conditioned on observed data. A related latent approach, hard
attention, fixes these issues, but is generally harder to train and less
accurate. This work considers variational attention networks, alternatives to
soft and hard attention for learning latent variable alignment models, with
tighter approximation bounds based on amortized variational inference. We
further propose methods for reducing the variance of gradients to make these
approaches computationally feasible. Experiments show that for machine
translation and visual question answering, inefficient exact latent variable
models outperform standard neural attention, but these gains go away when using
hard attention based training. On the other hand, variational attention retains
most of the performance gain but with training speed comparable to neural
attention.
|
Is Q-learning Provably Efficient?
|
Model-free reinforcement learning (RL) algorithms, such as Q-learning,
directly parameterize and update value functions or policies without explicitly
modeling the environment. They are typically simpler, more flexible to use, and
thus more prevalent in modern deep RL than model-based approaches. However,
empirical work has suggested that model-free algorithms may require more
samples to learn [Deisenroth and Rasmussen 2011, Schulman et al. 2015]. The
theoretical question of "whether model-free algorithms can be made sample
efficient" is one of the most fundamental questions in RL, and remains unsolved
even in the basic scenario with finitely many states and actions.
We prove that, in an episodic MDP setting, Q-learning with UCB exploration
achieves regret $\tilde{O}(\sqrt{H^3 SAT})$, where $S$ and $A$ are the numbers
of states and actions, $H$ is the number of steps per episode, and $T$ is the
total number of steps. This sample efficiency matches the optimal regret that
can be achieved by any model-based approach, up to a single $\sqrt{H}$ factor.
To the best of our knowledge, this is the first analysis in the model-free
setting that establishes $\sqrt{T}$ regret without requiring access to a
"simulator."
|
Kernel-Based Learning for Smart Inverter Control
|
Distribution grids are currently challenged by frequent voltage excursions
induced by intermittent solar generation. Smart inverters have been advocated
as a fast-responding means to regulate voltage and minimize ohmic losses. Since
optimal inverter coordination may be computationally challenging and preset
local control rules are subpar, the approach of customized control rules
designed in a quasi-static fashion features as a golden middle. Departing from
affine control rules, this work puts forth non-linear inverter control
policies. Drawing analogies to multi-task learning, reactive control is posed
as a kernel-based regression task. Leveraging a linearized grid model and given
anticipated data scenarios, inverter rules are jointly designed at the feeder
level to minimize a convex combination of voltage deviations and ohmic losses
via a linearly-constrained quadratic program. Numerical tests using real-world
data on a benchmark feeder demonstrate that nonlinear control rules driven also
by a few non-local readings can attain near-optimal performance.
|
Universal Transformers
|
Recurrent neural networks (RNNs) sequentially process data by updating their
state with each new data point, and have long been the de facto choice for
sequence modeling tasks. However, their inherently sequential computation makes
them slow to train. Feed-forward and convolutional architectures have recently
been shown to achieve superior results on some sequence modeling tasks such as
machine translation, with the added advantage that they concurrently process
all inputs in the sequence, leading to easy parallelization and faster training
times. Despite these successes, however, popular feed-forward sequence models
like the Transformer fail to generalize in many simple tasks that recurrent
models handle with ease, e.g. copying strings or even simple logical inference
when the string or formula lengths exceed those observed at training time. We
propose the Universal Transformer (UT), a parallel-in-time self-attentive
recurrent sequence model which can be cast as a generalization of the
Transformer model and which addresses these issues. UTs combine the
parallelizability and global receptive field of feed-forward sequence models
like the Transformer with the recurrent inductive bias of RNNs. We also add a
dynamic per-position halting mechanism and find that it improves accuracy on
several tasks. In contrast to the standard Transformer, under certain
assumptions, UTs can be shown to be Turing-complete. Our experiments show that
UTs outperform standard Transformers on a wide range of algorithmic and
language understanding tasks, including the challenging LAMBADA language
modeling task where UTs achieve a new state of the art, and machine translation
where UTs achieve a 0.9 BLEU improvement over Transformers on the WMT14 En-De
dataset.
|
Model-based free-breathing cardiac MRI reconstruction using deep learned
\& STORM priors: MoDL-STORM
|
We introduce a model-based reconstruction framework with deep learned (DL)
and smoothness regularization on manifolds (STORM) priors to recover free
breathing and ungated (FBU) cardiac MRI from highly undersampled measurements.
The DL priors enable us to exploit the local correlations, while the STORM
prior enables us to make use of the extensive non-local similarities that are
subject dependent. We introduce a novel model-based formulation that allows the
seamless integration of deep learning methods with available prior information,
which current deep learning algorithms are not capable of. The experimental
results demonstrate the preliminary potential of this work in accelerating FBU
cardiac MRI.
|
Algorithmic Framework for Model-based Deep Reinforcement Learning with
Theoretical Guarantees
|
Model-based reinforcement learning (RL) is considered to be a promising
approach to reduce the sample complexity that hinders model-free RL. However,
the theoretical understanding of such methods has been rather limited. This
paper introduces a novel algorithmic framework for designing and analyzing
model-based RL algorithms with theoretical guarantees. We design a
meta-algorithm with a theoretical guarantee of monotone improvement to a local
maximum of the expected reward. The meta-algorithm iteratively builds a lower
bound of the expected reward based on the estimated dynamical model and sample
trajectories, and then maximizes the lower bound jointly over the policy and
the model. The framework extends the optimism-in-face-of-uncertainty principle
to non-linear dynamical models in a way that requires \textit{no explicit}
uncertainty quantification. Instantiating our framework with simplification
gives a variant of model-based RL algorithms Stochastic Lower Bounds
Optimization (SLBO). Experiments demonstrate that SLBO achieves
state-of-the-art performance when only one million or fewer samples are
permitted on a range of continuous control benchmark tasks.
|
Learning Implicit Generative Models by Teaching Explicit Ones
|
Implicit generative models are difficult to train as no explicit density
functions are defined. Generative adversarial nets (GANs) present a minimax
framework to train such models, which however can suffer from mode collapse due
to the nature of the JS-divergence. This paper presents a learning by teaching
(LBT) approach to learning implicit models, which intrinsically avoids the mode
collapse problem by optimizing a KL-divergence rather than the JS-divergence in
GANs. In LBT, an auxiliary density estimator is introduced to fit the implicit
model's distribution while the implicit model teaches the density estimator to
match the data distribution. LBT is formulated as a bilevel optimization
problem, whose optimal generator matches the true data distribution. LBT can be
naturally integrated with GANs to derive a hybrid LBT-GAN that enjoys
complimentary benefits. Finally, we present a stochastic gradient ascent
algorithm with unrolling to solve the challenging learning problems.
Experimental results demonstrate the effectiveness of our method.
|
Automatic Gradient Boosting
|
Automatic machine learning performs predictive modeling with high performing
machine learning tools without human interference. This is achieved by making
machine learning applications parameter-free, i.e. only a dataset is provided
while the complete model selection and model building process is handled
internally through (often meta) optimization. Projects like Auto-WEKA and
auto-sklearn aim to solve the Combined Algorithm Selection and Hyperparameter
optimization (CASH) problem resulting in huge configuration spaces. However,
for most real-world applications, the optimization over only a few different
key learning algorithms can not only be sufficient, but also potentially
beneficial. The latter becomes apparent when one considers that models have to
be validated, explained, deployed and maintained. Here, less complex model are
often preferred, for validation or efficiency reasons, or even a strict
requirement. Automatic gradient boosting simplifies this idea one step further,
using only gradient boosting as a single learning algorithm in combination with
model-based hyperparameter tuning, threshold optimization and encoding of
categorical features. We introduce this general framework as well as a concrete
implementation called autoxgboost. It is compared to current AutoML projects on
16 datasets and despite its simplicity is able to achieve comparable results on
about half of the datasets as well as performing best on two.
|
Deep learning for comprehensive forecasting of Alzheimer's Disease
progression
|
Most approaches to machine learning from electronic health data can only
predict a single endpoint. Here, we present an alternative that uses
unsupervised deep learning to simulate detailed patient trajectories. We use
data comprising 18-month trajectories of 44 clinical variables from 1908
patients with Mild Cognitive Impairment or Alzheimer's Disease to train a model
for personalized forecasting of disease progression. We simulate synthetic
patient data including the evolution of each sub-component of cognitive exams,
laboratory tests, and their associations with baseline clinical
characteristics, generating both predictions and their confidence intervals.
Our unsupervised model predicts changes in total ADAS-Cog scores with the same
accuracy as specifically trained supervised models and identifies
sub-components associated with word recall as predictive of progression. The
ability to simultaneously simulate dozens of patient characteristics is a
crucial step towards personalized medicine for Alzheimer's Disease.
|
Deep Structured Generative Models
|
Deep generative models have shown promising results in generating realistic
images, but it is still non-trivial to generate images with complicated
structures. The main reason is that most of the current generative models fail
to explore the structures in the images including spatial layout and semantic
relations between objects. To address this issue, we propose a novel deep
structured generative model which boosts generative adversarial networks (GANs)
with the aid of structure information. In particular, the layout or structure
of the scene is encoded by a stochastic and-or graph (sAOG), in which the
terminal nodes represent single objects and edges represent relations between
objects. With the sAOG appropriately harnessed, our model can successfully
capture the intrinsic structure in the scenes and generate images of
complicated scenes accordingly. Furthermore, a detection network is introduced
to infer scene structures from a image. Experimental results demonstrate the
effectiveness of our proposed method on both modeling the intrinsic structures,
and generating realistic images.
|
DeepDiff: Deep-learning for predicting Differential gene expression from
histone modifications
|
Computational methods that predict differential gene expression from histone
modification signals are highly desirable for understanding how histone
modifications control the functional heterogeneity of cells through influencing
differential gene regulation. Recent studies either failed to capture
combinatorial effects on differential prediction or primarily only focused on
cell type-specific analysis. In this paper, we develop a novel attention-based
deep learning architecture, DeepDiff, that provides a unified and end-to-end
solution to model and to interpret how dependencies among histone modifications
control the differential patterns of gene regulation. DeepDiff uses a hierarchy
of multiple Long short-term memory (LSTM) modules to encode the spatial
structure of input signals and to model how various histone modifications
cooperate automatically. We introduce and train two levels of attention jointly
with the target prediction, enabling DeepDiff to attend differentially to
relevant modifications and to locate important genome positions for each
modification. Additionally, DeepDiff introduces a novel deep-learning based
multi-task formulation to use the cell-type-specific gene expression
predictions as auxiliary tasks, encouraging richer feature embeddings in our
primary task of differential expression prediction. Using data from Roadmap
Epigenomics Project (REMC) for ten different pairs of cell types, we show that
DeepDiff significantly outperforms the state-of-the-art baselines for
differential gene expression prediction. The learned attention weights are
validated by observations from previous studies about how epigenetic mechanisms
connect to differential gene expression. Codes and results are available at
\url{deepchrome.org}
|
A Simple Unified Framework for Detecting Out-of-Distribution Samples and
Adversarial Attacks
|
Detecting test samples drawn sufficiently far away from the training
distribution statistically or adversarially is a fundamental requirement for
deploying a good classifier in many real-world machine learning applications.
However, deep neural networks with the softmax classifier are known to produce
highly overconfident posterior distributions even for such abnormal samples. In
this paper, we propose a simple yet effective method for detecting any abnormal
samples, which is applicable to any pre-trained softmax neural classifier. We
obtain the class conditional Gaussian distributions with respect to (low- and
upper-level) features of the deep models under Gaussian discriminant analysis,
which result in a confidence score based on the Mahalanobis distance. While
most prior methods have been evaluated for detecting either out-of-distribution
or adversarial samples, but not both, the proposed method achieves the
state-of-the-art performances for both cases in our experiments. Moreover, we
found that our proposed method is more robust in harsh cases, e.g., when the
training dataset has noisy labels or small number of samples. Finally, we show
that the proposed method enjoys broader usage by applying it to
class-incremental learning: whenever out-of-distribution samples are detected,
our classification rule can incorporate new classes well without further
training deep models.
|
The Limit Points of (Optimistic) Gradient Descent in Min-Max
Optimization
|
Motivated by applications in Optimization, Game Theory, and the training of
Generative Adversarial Networks, the convergence properties of first order
methods in min-max problems have received extensive study. It has been
recognized that they may cycle, and there is no good understanding of their
limit points when they do not. When they converge, do they converge to local
min-max solutions? We characterize the limit points of two basic first order
methods, namely Gradient Descent/Ascent (GDA) and Optimistic Gradient Descent
Ascent (OGDA). We show that both dynamics avoid unstable critical points for
almost all initializations. Moreover, for small step sizes and under mild
assumptions, the set of \{OGDA\}-stable critical points is a superset of
\{GDA\}-stable critical points, which is a superset of local min-max solutions
(strict in some cases). The connecting thread is that the behavior of these
dynamics can be studied from a dynamical systems perspective.
|
Emotion Recognition from Speech based on Relevant Feature and Majority
Voting
|
This paper proposes an approach to detect emotion from human speech employing
majority voting technique over several machine learning techniques. The
contribution of this work is in two folds: firstly it selects those features of
speech which is most promising for classification and secondly it uses the
majority voting technique that selects the exact class of emotion. Here,
majority voting technique has been applied over Neural Network (NN), Decision
Tree (DT), Support Vector Machine (SVM) and K-Nearest Neighbor (KNN). Input
vector of NN, DT, SVM and KNN consists of various acoustic and prosodic
features like Pitch, Mel-Frequency Cepstral coefficients etc. From speech
signal many feature have been extracted and only promising features have been
selected. To consider a feature as promising, Fast Correlation based feature
selection (FCBF) and Fisher score algorithms have been used and only those
features are selected which are highly ranked by both of them. The proposed
approach has been tested on Berlin dataset of emotional speech [3] and
Electromagnetic Articulography (EMA) dataset [4]. The experimental result shows
that majority voting technique attains better accuracy over individual machine
learning techniques. The employment of the proposed approach can effectively
recognize the emotion of human beings in case of social robot, intelligent chat
client, call-center of a company etc.
|
Seq2Seq2Sentiment: Multimodal Sequence to Sequence Models for Sentiment
Analysis
|
Multimodal machine learning is a core research area spanning the language,
visual and acoustic modalities. The central challenge in multimodal learning
involves learning representations that can process and relate information from
multiple modalities. In this paper, we propose two methods for unsupervised
learning of joint multimodal representations using sequence to sequence
(Seq2Seq) methods: a \textit{Seq2Seq Modality Translation Model} and a
\textit{Hierarchical Seq2Seq Modality Translation Model}. We also explore
multiple different variations on the multimodal inputs and outputs of these
seq2seq models. Our experiments on multimodal sentiment analysis using the
CMU-MOSI dataset indicate that our methods learn informative multimodal
representations that outperform the baselines and achieve improved performance
on multimodal sentiment analysis, specifically in the Bimodal case where our
model is able to improve F1 Score by twelve points. We also discuss future
directions for multimodal Seq2Seq methods.
|
Discovering Interesting Plots in Production Yield Data Analytics
|
An analytic process is iterative between two agents, an analyst and an
analytic toolbox. Each iteration comprises three main steps: preparing a
dataset, running an analytic tool, and evaluating the result, where dataset
preparation and result evaluation, conducted by the analyst, are largely
domain-knowledge driven. In this work, the focus is on automating the result
evaluation step. The underlying problem is to identify plots that are deemed
interesting by an analyst. We propose a methodology to learn such analyst's
intent based on Generative Adversarial Networks (GANs) and demonstrate its
applications in the context of production yield optimization using data
collected from several product lines.
|
Quantification under prior probability shift: the ratio estimator and
its extensions
|
The quantification problem consists of determining the prevalence of a given
label in a target population. However, one often has access to the labels in a
sample from the training population but not in the target population. A common
assumption in this situation is that of prior probability shift, that is, once
the labels are known, the distribution of the features is the same in the
training and target populations. In this paper, we derive a new lower bound for
the risk of the quantification problem under the prior shift assumption.
Complementing this lower bound, we present a new approximately minimax class of
estimators, ratio estimators, which generalize several previous proposals in
the literature. Using a weaker version of the prior shift assumption, which can
be tested, we show that ratio estimators can be used to build confidence
intervals for the quantification problem. We also extend the ratio estimator so
that it can: (i) incorporate labels from the target population, when they are
available and (ii) estimate how the prevalence of positive labels varies
according to a function of certain covariates.
|
A Hierarchical Bayesian Linear Regression Model with Local Features for
Stochastic Dynamics Approximation
|
One of the challenges in model-based control of stochastic dynamical systems
is that the state transition dynamics are involved, and it is not easy or
efficient to make good-quality predictions of the states. Moreover, there are
not many representational models for the majority of autonomous systems, as it
is not easy to build a compact model that captures the entire dynamical
subtleties and uncertainties. In this work, we present a hierarchical Bayesian
linear regression model with local features to learn the dynamics of a
micro-robotic system as well as two simpler examples, consisting of a
stochastic mass-spring damper and a stochastic double inverted pendulum on a
cart. The model is hierarchical since we assume non-stationary priors for the
model parameters. These non-stationary priors make the model more flexible by
imposing priors on the priors of the model. To solve the maximum likelihood
(ML) problem for this hierarchical model, we use the variational expectation
maximization (EM) algorithm, and enhance the procedure by introducing hidden
target variables. The algorithm yields parsimonious model structures, and
consistently provides fast and accurate predictions for all our examples
involving large training and test sets. This demonstrates the effectiveness of
the method in learning stochastic dynamics, which makes it suitable for future
use in a paradigm, such as model-based reinforcement learning, to compute
optimal control policies in real time.
|
Instance-based entropy fuzzy support vector machine for imbalanced data
|
Imbalanced classification has been a major challenge for machine learning
because many standard classifiers mainly focus on balanced datasets and tend to
have biased results towards the majority class. We modify entropy fuzzy support
vector machine (EFSVM) and introduce instance-based entropy fuzzy support
vector machine (IEFSVM). Both EFSVM and IEFSVM use the entropy information of
k-nearest neighbors to determine the fuzzy membership value for each sample
which prioritizes the importance of each sample. IEFSVM considers the diversity
of entropy patterns for each sample when increasing the size of neighbors, k,
while EFSVM uses single entropy information of the fixed size of neighbors for
all samples. By varying k, we can reflect the component change of sample's
neighbors from near to far distance in the determination of fuzzy value
membership. Numerical experiments on 35 public and 12 real-world imbalanced
datasets are performed to validate IEFSVM and area under the receiver operating
characteristic curve (AUC) is used to compare its performance with other SVMs
and machine learning methods. IEFSVM shows a much higher AUC value for datasets
with high imbalance ratio, implying that IEFSVM is effective in dealing with
the class imbalance problem.
|
Adaptive Learning Method of Recurrent Temporal Deep Belief Network to
Analyze Time Series Data
|
Deep Learning has the hierarchical network architecture to represent the
complicated features of input patterns. Such architecture is well known to
represent higher learning capability compared with some conventional models if
the best set of parameters in the optimal network structure is found. We have
been developing the adaptive learning method that can discover the optimal
network structure in Deep Belief Network (DBN). The learning method can
construct the network structure with the optimal number of hidden neurons in
each Restricted Boltzmann Machine and with the optimal number of layers in the
DBN during learning phase. The network structure of the learning method can be
self-organized according to given input patterns of big data set. In this
paper, we embed the adaptive learning method into the recurrent temporal RBM
and the self-generated layer into DBN. In order to verify the effectiveness of
our proposed method, the experimental results are higher classification
capability than the conventional methods in this paper.
|
Learning Neural Models for End-to-End Clustering
|
We propose a novel end-to-end neural network architecture that, once trained,
directly outputs a probabilistic clustering of a batch of input examples in one
pass. It estimates a distribution over the number of clusters $k$, and for each
$1 \leq k \leq k_\mathrm{max}$, a distribution over the individual cluster
assignment for each data point. The network is trained in advance in a
supervised fashion on separate data to learn grouping by any perceptual
similarity criterion based on pairwise labels (same/different group). It can
then be applied to different data containing different groups. We demonstrate
promising performance on high-dimensional data like images (COIL-100) and
speech (TIMIT). We call this ``learning to cluster'' and show its conceptual
difference to deep metric learning, semi-supervise clustering and other related
approaches while having the advantage of performing learnable clustering fully
end-to-end.
|
Causal Discovery in the Presence of Missing Data
|
Missing data are ubiquitous in many domains including healthcare. When these
data entries are not missing completely at random, the (conditional)
independence relations in the observed data may be different from those in the
complete data generated by the underlying causal process. Consequently, simply
applying existing causal discovery methods to the observed data may lead to
wrong conclusions. In this paper, we aim at developing a causal discovery
method to recover the underlying causal structure from observed data that
follow different missingness mechanisms, including missing completely at random
(MCAR), missing at random (MAR), and missing not at random (MNAR). With
missingness mechanisms represented by missingness graphs, we analyse conditions
under which additional correction is needed to derive conditional
independence/dependence relations in the complete data. Based on our analysis,
we propose the Missing Value PC (MVPC) algorithm for both continuous and binary
variables, which extends the PC algorithm to incorporate additional
corrections. Our proposed MVPC is shown in theory to give asymptotically
correct results even on data that are MAR or MNAR. Experimental results on
synthetic data show that the proposed algorithm is able to find correct causal
relations even in the general case of MNAR. Moreover, we create a neuropathic
pain diagnostic simulator for evaluating causal discovery methods. Evaluated on
such simulated neuropathic pain diagnosis records and the other two real world
applications, MVPC outperforms the other benchmark methods.
|
Medusa: A Scalable Interconnect for Many-Port DNN Accelerators and Wide
DRAM Controller Interfaces
|
To cope with the increasing demand and computational intensity of deep neural
networks (DNNs), industry and academia have turned to accelerator technologies.
In particular, FPGAs have been shown to provide a good balance between
performance and energy efficiency for accelerating DNNs. While significant
research has focused on how to build efficient layer processors, the
computational building blocks of DNN accelerators, relatively little attention
has been paid to the on-chip interconnects that sit between the layer
processors and the FPGA's DRAM controller.
We observe a disparity between DNN accelerator interfaces, which tend to
comprise many narrow ports, and FPGA DRAM controller interfaces, which tend to
be wide buses. This mismatch causes traditional interconnects to consume
significant FPGA resources. To address this problem, we designed Medusa: an
optimized FPGA memory interconnect which transposes data in the interconnect
fabric, tailoring the interconnect to the needs of DNN layer processors.
Compared to a traditional FPGA interconnect, our design can reduce LUT and FF
use by 4.7x and 6.0x, and improves frequency by 1.8x.
|
On Catastrophic Forgetting and Mode Collapse in Generative Adversarial
Networks
|
In this paper, we show that Generative Adversarial Networks (GANs) suffer
from catastrophic forgetting even when they are trained to approximate a single
target distribution. We show that GAN training is a continual learning problem
in which the sequence of changing model distributions is the sequence of tasks
to the discriminator. The level of mismatch between tasks in the sequence
determines the level of forgetting. Catastrophic forgetting is interrelated to
mode collapse and can make the training of GANs non-convergent. We investigate
the landscape of the discriminator's output in different variants of GANs and
find that when a GAN converges to a good equilibrium, real training datapoints
are wide local maxima of the discriminator. We empirically show the
relationship between the sharpness of local maxima and mode collapse and
generalization in GANs. We show how catastrophic forgetting prevents the
discriminator from making real datapoints local maxima, and thus causes
non-convergence. Finally, we study methods for preventing catastrophic
forgetting in GANs.
|
Improved SVD-based Initialization for Nonnegative Matrix Factorization
using Low-Rank Correction
|
Due to the iterative nature of most nonnegative matrix factorization
(\textsc{NMF}) algorithms, initialization is a key aspect as it significantly
influences both the convergence and the final solution obtained. Many
initialization schemes have been proposed for NMF, among which one of the most
popular class of methods are based on the singular value decomposition (SVD).
However, these SVD-based initializations do not satisfy a rather natural
condition, namely that the error should decrease as the rank of factorization
increases. In this paper, we propose a novel SVD-based \textsc{NMF}
initialization to specifically address this shortcoming by taking into account
the SVD factors that were discarded to obtain a nonnegative initialization.
This method, referred to as nonnegative SVD with low-rank correction
(NNSVD-LRC), allows us to significantly reduce the initial error at a
negligible additional computational cost using the low-rank structure of the
discarded SVD factors. NNSVD-LRC has two other advantages compared to previous
SVD-based initializations: (1) it provably generates sparse initial factors,
and (2) it is faster as it only requires to compute a truncated SVD of rank
$\lceil r/2 + 1 \rceil$ where $r$ is the factorization rank of the sought NMF
decomposition (as opposed to a rank-$r$ truncated SVD for other methods). We
show on several standard dense and sparse data sets that our new method
competes favorably with state-of-the-art SVD-based initializations for NMF.
|
Learning Singularity Avoidance
|
With the increase in complexity of robotic systems and the rise in non-expert
users, it can be assumed that task constraints are not explicitly known. In
tasks where avoiding singularity is critical to its success, this paper
provides an approach, especially for non-expert users, for the system to learn
the constraints contained in a set of demonstrations, such that they can be
used to optimise an autonomous controller to avoid singularity, without having
to explicitly know the task constraints. The proposed approach avoids
singularity, and thereby unpredictable behaviour when carrying out a task, by
maximising the learnt manipulability throughout the motion of the constrained
system, and is not limited to kinematic systems. Its benefits are demonstrated
through comparisons with other control policies which show that the constrained
manipulability of a system learnt through demonstration can be used to avoid
singularities in cases where these other policies would fail. In the absence of
the systems manipulability subject to a tasks constraints, the proposed
approach can be used instead to infer these with results showing errors less
than 10^-5 in 3DOF simulated systems as well as 10^-2 using a 7DOF real world
robotic system.
|
Temporal Convolution Networks for Real-Time Abdominal Fetal Aorta
Analysis with Ultrasound
|
The automatic analysis of ultrasound sequences can substantially improve the
efficiency of clinical diagnosis. In this work we present our attempt to
automate the challenging task of measuring the vascular diameter of the fetal
abdominal aorta from ultrasound images. We propose a neural network
architecture consisting of three blocks: a convolutional layer for the
extraction of imaging features, a Convolution Gated Recurrent Unit (C-GRU) for
enforcing the temporal coherence across video frames and exploiting the
temporal redundancy of a signal, and a regularized loss function, called
\textit{CyclicLoss}, to impose our prior knowledge about the periodicity of the
observed signal. We present experimental evidence suggesting that the proposed
architecture can reach an accuracy substantially superior to previously
proposed methods, providing an average reduction of the mean squared error from
$0.31 mm^2$ (state-of-art) to $0.09 mm^2$, and a relative error reduction from
$8.1\%$ to $5.3\%$. The mean execution speed of the proposed approach of 289
frames per second makes it suitable for real time clinical use.
|
Recurrent Neural Networks with Flexible Gates using Kernel Activation
Functions
|
Gated recurrent neural networks have achieved remarkable results in the
analysis of sequential data. Inside these networks, gates are used to control
the flow of information, allowing to model even very long-term dependencies in
the data. In this paper, we investigate whether the original gate equation (a
linear projection followed by an element-wise sigmoid) can be improved. In
particular, we design a more flexible architecture, with a small number of
adaptable parameters, which is able to model a wider range of gating functions
than the classical one. To this end, we replace the sigmoid function in the
standard gate with a non-parametric formulation extending the recently proposed
kernel activation function (KAF), with the addition of a residual
skip-connection. A set of experiments on sequential variants of the MNIST
dataset shows that the adoption of this novel gate allows to improve accuracy
with a negligible cost in terms of computational power and with a large
speed-up in the number of training iterations.
|
A punishment voting algorithm based on super categories construction for
acoustic scene classification
|
In acoustic scene classification researches, audio segment is usually split
into multiple samples. Majority voting is then utilized to ensemble the results
of the samples. In this paper, we propose a punishment voting algorithm based
on the super categories construction method for acoustic scene classification.
Specifically, we propose a DenseNet-like model as the base classifier. The base
classifier is trained by the CQT spectrograms generated from the raw audio
segments. Taking advantage of the results of the base classifier, we propose a
super categories construction method using the spectral clustering. Super
classifiers corresponding to the constructed super categories are further
trained. Finally, the super classifiers are utilized to enhance the majority
voting of the base classifier by punishment voting. Experiments show that the
punishment voting obviously improves the performances on both the DCASE2017
Development dataset and the LITIS Rouen dataset.
|
Proactive Intervention to Downtrend Employee Attrition using Artificial
Intelligence Techniques
|
To predict the employee attrition beforehand and to enable management to take
individualized preventive action. Using Ensemble classification modeling
techniques and Linear Regression. Model could predict over 91% accurate
employee prediction, lead-time in separation and individual reasons causing
attrition. Prior intimation of employee attrition enables manager to take
preventive actions to retain employee or to manage the business consequences of
attrition. Once deployed this will model can help in downtrend Employee
Attrition, will help manager to manage team more effectively. Model does not
cover the natural calamities, and unforeseen events occurring at an individual
level like accident, death etc.
|
FINN-L: Library Extensions and Design Trade-off Analysis for Variable
Precision LSTM Networks on FPGAs
|
It is well known that many types of artificial neural networks, including
recurrent networks, can achieve a high classification accuracy even with
low-precision weights and activations. The reduction in precision generally
yields much more efficient hardware implementations in regards to hardware
cost, memory requirements, energy, and achievable throughput. In this paper, we
present the first systematic exploration of this design space as a function of
precision for Bidirectional Long Short-Term Memory (BiLSTM) neural network.
Specifically, we include an in-depth investigation of precision vs. accuracy
using a fully hardware-aware training flow, where during training quantization
of all aspects of the network including weights, input, output and in-memory
cell activations are taken into consideration. In addition, hardware resource
cost, power consumption and throughput scalability are explored as a function
of precision for FPGA-based implementations of BiLSTM, and multiple approaches
of parallelizing the hardware. We provide the first open source HLS library
extension of FINN for parameterizable hardware architectures of LSTM layers on
FPGAs which offers full precision flexibility and allows for parameterizable
performance scaling offering different levels of parallelism within the
architecture. Based on this library, we present an FPGA-based accelerator for
BiLSTM neural network designed for optical character recognition, along with
numerous other experimental proof points for a Zynq UltraScale+ XCZU7EV MPSoC
within the given design space.
|
A Recurrent Neural Network Survival Model: Predicting Web User Return
Time
|
The size of a website's active user base directly affects its value. Thus, it
is important to monitor and influence a user's likelihood to return to a site.
Essential to this is predicting when a user will return. Current state of the
art approaches to solve this problem come in two flavors: (1) Recurrent Neural
Network (RNN) based solutions and (2) survival analysis methods. We observe
that both techniques are severely limited when applied to this problem.
Survival models can only incorporate aggregate representations of users instead
of automatically learning a representation directly from a raw time series of
user actions. RNNs can automatically learn features, but can not be directly
trained with examples of non-returning users who have no target value for their
return time. We develop a novel RNN survival model that removes the limitations
of the state of the art methods. We demonstrate that this model can
successfully be applied to return time prediction on a large e-commerce dataset
with a superior ability to discriminate between returning and non-returning
users than either method applied in isolation.
|
VFunc: a Deep Generative Model for Functions
|
We introduce a deep generative model for functions. Our model provides a
joint distribution p(f, z) over functions f and latent variables z which lets
us efficiently sample from the marginal p(f) and maximize a variational lower
bound on the entropy H(f). We can thus maximize objectives of the form
E_{f~p(f)}[R(f)] + c*H(f), where R(f) denotes, e.g., a data log-likelihood term
or an expected reward. Such objectives encompass Bayesian deep learning in
function space, rather than parameter space, and Bayesian deep RL with
representations of uncertainty that offer benefits over bootstrapping and
parameter noise. In this short paper we describe our model, situate it in the
context of prior work, and present proof-of-concept experiments for regression
and RL.
|
Modeling and Soft-fault Diagnosis of Underwater Thrusters with Recurrent
Neural Networks
|
Noncritical soft-faults and model deviations are a challenge for Fault
Detection and Diagnosis (FDD) of resident Autonomous Underwater Vehicles
(AUVs). Such systems may have a faster performance degradation due to the
permanent exposure to the marine environment, and constant monitoring of
component conditions is required to ensure their reliability. This works
presents an evaluation of Recurrent Neural Networks (RNNs) for a data-driven
fault detection and diagnosis scheme for underwater thrusters with empirical
data. The nominal behavior of the thruster was modeled using the measured
control input, voltage, rotational speed and current signals. We evaluated the
performance of fault classification using all the measured signals compared to
using the computed residuals from the nominal model as features.
|
Exploiting statistical dependencies of time series with hierarchical
correlation reconstruction
|
While we are usually focused on forecasting future values of time series, it
is often valuable to additionally predict their entire probability
distributions, e.g. to evaluate risk, Monte Carlo simulations. On example of
time series of $\approx$ 30000 Dow Jones Industrial Averages, there will be
presented application of hierarchical correlation reconstruction for this
purpose: MSE estimating polynomial as joint density for (current value,
context), where context is for example a few previous values. Then substituting
the currently observed context and normalizing density to 1, we get predicted
probability distribution for the current value. In contrast to standard machine
learning approaches like neural networks, optimal polynomial coefficients here
have inexpensive direct formula, have controllable accuracy, are unique and
independently calculated, each has a specific cumulant-like interpretation, and
such approximation can asymptotically approach complete description of any real
joint distribution - providing universal tool to quantitatively describe and
exploit statistical dependencies in time series, systematically enhancing
ARMA/ARCH-like approaches, also based on different distributions than Gaussian
which turns out improper for daily log returns. There is also discussed
application for non-stationary time series like calculating linear time trend,
or adapting coefficients to local statistical behavior.
|
TherML: Thermodynamics of Machine Learning
|
In this work we offer a framework for reasoning about a wide class of
existing objectives in machine learning. We develop a formal correspondence
between this work and thermodynamics and discuss its implications.
|
Optimization over Continuous and Multi-dimensional Decisions with
Observational Data
|
We consider the optimization of an uncertain objective over continuous and
multi-dimensional decision spaces in problems in which we are only provided
with observational data. We propose a novel algorithmic framework that is
tractable, asymptotically consistent, and superior to comparable methods on
example problems. Our approach leverages predictive machine learning methods
and incorporates information on the uncertainty of the predicted outcomes for
the purpose of prescribing decisions. We demonstrate the efficacy of our method
on examples involving both synthetic and real data sets.
|
A Hardware-Software Blueprint for Flexible Deep Learning Specialization
|
Specialized Deep Learning (DL) acceleration stacks, designed for a specific
set of frameworks, model architectures, operators, and data types, offer the
allure of high performance while sacrificing flexibility. Changes in
algorithms, models, operators, or numerical systems threaten the viability of
specialized hardware accelerators. We propose VTA, a programmable deep learning
architecture template designed to be extensible in the face of evolving
workloads. VTA achieves this flexibility via a parametrizable architecture,
two-level ISA, and a JIT compiler. The two-level ISA is based on (1) a task-ISA
that explicitly orchestrates concurrent compute and memory tasks and (2) a
microcode-ISA which implements a wide variety of operators with single-cycle
tensor-tensor operations. Next, we propose a runtime system equipped with a JIT
compiler for flexible code-generation and heterogeneous execution that enables
effective use of the VTA architecture. VTA is integrated and open-sourced into
Apache TVM, a state-of-the-art deep learning compilation stack that provides
flexibility for diverse models and divergent hardware backends. We propose a
flow that performs design space exploration to generate a customized hardware
architecture and software operator library that can be leveraged by mainstream
learning frameworks. We demonstrate our approach by deploying optimized deep
learning models used for object classification and style transfer on edge-class
FPGAs.
|
Distributed Variational Representation Learning
|
The problem of distributed representation learning is one in which multiple
sources of information $X_1,\ldots,X_K$ are processed separately so as to learn
as much information as possible about some ground truth $Y$. We investigate
this problem from information-theoretic grounds, through a generalization of
Tishby's centralized Information Bottleneck (IB) method to the distributed
setting. Specifically, $K$ encoders, $K \geq 2$, compress their observations
$X_1,\ldots,X_K$ separately in a manner such that, collectively, the produced
representations preserve as much information as possible about $Y$. We study
both discrete memoryless (DM) and memoryless vector Gaussian data models. For
the discrete model, we establish a single-letter characterization of the
optimal tradeoff between complexity (or rate) and relevance (or information)
for a class of memoryless sources (the observations $X_1,\ldots,X_K$ being
conditionally independent given $Y$). For the vector Gaussian model, we provide
an explicit characterization of the optimal complexity-relevance tradeoff.
Furthermore, we develop a variational bound on the complexity-relevance
tradeoff which generalizes the evidence lower bound (ELBO) to the distributed
setting. We also provide two algorithms that allow to compute this bound: i) a
Blahut-Arimoto type iterative algorithm which enables to compute optimal
complexity-relevance encoding mappings by iterating over a set of
self-consistent equations, and ii) a variational inference type algorithm in
which the encoding mappings are parametrized by neural networks and the bound
approximated by Markov sampling and optimized with stochastic gradient descent.
Numerical results on synthetic and real datasets are provided to support the
efficiency of the approaches and algorithms developed in this paper.
|
Differentially Private False Discovery Rate Control
|
Differential privacy provides a rigorous framework for privacy-preserving
data analysis. This paper proposes the first differentially private procedure
for controlling the false discovery rate (FDR) in multiple hypothesis testing.
Inspired by the Benjamini-Hochberg procedure (BHq), our approach is to first
repeatedly add noise to the logarithms of the $p$-values to ensure differential
privacy and to select an approximately smallest $p$-value serving as a
promising candidate at each iteration; the selected $p$-values are further
supplied to the BHq and our private procedure releases only the rejected ones.
Moreover, we develop a new technique that is based on a backward submartingale
for proving FDR control of a broad class of multiple testing procedures,
including our private procedure, and both the BHq step-up and step-down
procedures. As a novel aspect, the proof works for arbitrary dependence between
the true null and false null test statistics, while FDR control is maintained
up to a small multiplicative factor.
|
Make $\ell_1$ Regularization Effective in Training Sparse CNN
|
Compressed Sensing using $\ell_1$ regularization is among the most powerful
and popular sparsification technique in many applications, but why has it not
been used to obtain sparse deep learning model such as convolutional neural
network (CNN)? This paper is aimed to provide an answer to this question and to
show how to make it work. We first demonstrate that the commonly used
stochastic gradient decent (SGD) and variants training algorithm is not an
appropriate match with $\ell_1$ regularization and then replace it with a
different training algorithm based on a regularized dual averaging (RDA)
method. RDA was originally designed specifically for convex problem, but with
new theoretical insight and algorithmic modifications (using proper
initialization and adaptivity), we have made it an effective match with
$\ell_1$ regularization to achieve a state-of-the-art sparsity for CNN compared
to other weight pruning methods without compromising accuracy (achieving 95\%
sparsity for ResNet18 on CIFAR-10, for example).
|
Measuring abstract reasoning in neural networks
|
Whether neural networks can learn abstract reasoning or whether they merely
rely on superficial statistics is a topic of recent debate. Here, we propose a
dataset and challenge designed to probe abstract reasoning, inspired by a
well-known human IQ test. To succeed at this challenge, models must cope with
various generalisation `regimes' in which the training and test data differ in
clearly-defined ways. We show that popular models such as ResNets perform
poorly, even when the training and test sets differ only minimally, and we
present a novel architecture, with a structure designed to encourage reasoning,
that does significantly better. When we vary the way in which the test
questions and training data differ, we find that our model is notably
proficient at certain forms of generalisation, but notably weak at others. We
further show that the model's ability to generalise improves markedly if it is
trained to predict symbolic explanations for its answers. Altogether, we
introduce and explore ways to both measure and induce stronger abstract
reasoning in neural networks. Our freely-available dataset should motivate
further progress in this direction.
|
Morse Code Datasets for Machine Learning
|
We present an algorithm to generate synthetic datasets of tunable difficulty
on classification of Morse code symbols for supervised machine learning
problems, in particular, neural networks. The datasets are spatially
one-dimensional and have a small number of input features, leading to high
density of input information content. This makes them particularly challenging
when implementing network complexity reduction methods. We explore how network
performance is affected by deliberately adding various forms of noise and
expanding the feature set and dataset size. Finally, we establish several
metrics to indicate the difficulty of a dataset, and evaluate their merits. The
algorithm and datasets are open-source.
|
DeepMove: Learning Place Representations through Large Scale Movement
Data
|
Understanding and reasoning about places and their relationships are critical
for many applications. Places are traditionally curated by a small group of
people as place gazetteers and are represented by an ID with spatial extent,
category, and other descriptions. However, a place context is described to a
large extent by movements made from/to other places. Places are linked and
related to each other by these movements. This important context is missing
from the traditional representation.
We present DeepMove, a novel approach for learning latent representations of
places. DeepMove advances the current deep learning based place representations
by directly model movements between places. We demonstrate DeepMove's latent
representations on place categorization and clustering tasks on large place and
movement datasets with respect to important parameters. Our results show that
DeepMove outperforms state-of-the-art baselines. DeepMove's representations can
provide up to 15% higher than competing methods in matching rate of place
category and result in up to 39% higher silhouette coefficient value for place
clusters.
DeepMove is spatial and temporal context aware. It is scalable. It
outperforms competing models using much smaller training dataset (a month or
1/12 of data). These qualities make it suitable for a broad class of real-world
applications.
|
Estimating Cellular Goals from High-Dimensional Biological Data
|
Optimization-based models have been used to predict cellular behavior for
over 25 years. The constraints in these models are derived from genome
annotations, measured macro-molecular composition of cells, and by measuring
the cell's growth rate and metabolism in different conditions. The cellular
goal (the optimization problem that the cell is trying to solve) can be
challenging to derive experimentally for many organisms, including human or
mammalian cells, which have complex metabolic capabilities and are not well
understood. Existing approaches to learning goals from data include (a)
estimating a linear objective function, or (b) estimating linear constraints
that model complex biochemical reactions and constrain the cell's operation.
The latter approach is important because often the known/observed biochemical
reactions are not enough to explain observations, and hence there is a need to
extend automatically the model complexity by learning new chemical reactions.
However, this leads to nonconvex optimization problems, and existing tools
cannot scale to realistically large metabolic models. Hence, constraint
estimation is still used sparingly despite its benefits for modeling cell
metabolism, which is important for developing novel antimicrobials against
pathogens, discovering cancer drug targets, and producing value-added
chemicals. Here, we develop the first approach to estimating constraint
reactions from data that can scale to realistically large metabolic models.
Previous tools have been used on problems having less than 75 biochemical
reactions and 60 metabolites, which limits real-life-size applications. We
perform extensive experiments using 75 large-scale metabolic network models for
different organisms (including bacteria, yeasts, and mammals) and show that our
algorithm can recover cellular constraint reactions, even when some
measurements are missing.
|
On the Fundamental Limits of Coded Data Shuffling for Distributed
Machine Learning
|
We consider the data shuffling problem in a distributed learning system, in
which a master node is connected to a set of worker nodes, via a shared link,
in order to communicate a set of files to the worker nodes. The master node has
access to a database of files. In every shuffling iteration, each worker node
processes a new subset of files, and has excess storage to partially cache the
remaining files, assuming the cached files are uncoded. The caches of the
worker nodes are updated every iteration, and they should be designed to
satisfy any possible unknown permutation of the files in subsequent iterations.
For this problem, we characterize the exact load-memory trade-off for
worst-case shuffling by deriving the minimum communication load for a given
storage capacity per worker node. As a byproduct, the exact load-memory
trade-off for any shuffling is characterized when the number of files is equal
to the number of worker nodes. We propose a novel deterministic coded shuffling
scheme, which improves the state of the art, by exploiting the cache memories
to create coded functions that can be decoded by several worker nodes. Then, we
prove the optimality of our proposed scheme by deriving a matching lower bound
and showing that the placement phase of the proposed coded shuffling scheme is
optimal over all shuffles.
|
Phase Retrieval Under a Generative Prior
|
The phase retrieval problem asks to recover a natural signal $y_0 \in
\mathbb{R}^n$ from $m$ quadratic observations, where $m$ is to be minimized. As
is common in many imaging problems, natural signals are considered sparse with
respect to a known basis, and the generic sparsity prior is enforced via
$\ell_1$ regularization. While successful in the realm of linear inverse
problems, such $\ell_1$ methods have encountered possibly fundamental
limitations, as no computationally efficient algorithm for phase retrieval of a
$k$-sparse signal has been proven to succeed with fewer than $O(k^2\log n)$
generic measurements, exceeding the theoretical optimum of $O(k \log n)$. In
this paper, we propose a novel framework for phase retrieval by 1) modeling
natural signals as being in the range of a deep generative neural network $G :
\mathbb{R}^k \rightarrow \mathbb{R}^n$ and 2) enforcing this prior directly by
optimizing an empirical risk objective over the domain of the generator. Our
formulation has provably favorable global geometry for gradient methods, as
soon as $m = O(kd^2\log n)$, where $d$ is the depth of the network.
Specifically, when suitable deterministic conditions on the generator and
measurement matrix are met, we construct a descent direction for any point
outside of a small neighborhood around the unique global minimizer and its
negative multiple, and show that such conditions hold with high probability
under Gaussian ensembles of multilayer fully-connected generator networks and
measurement matrices. This formulation for structured phase retrieval thus has
two advantages over sparsity based methods: 1) deep generative priors can more
tightly represent natural signals and 2) information theoretically optimal
sample complexity. We corroborate these results with experiments showing that
exploiting generative models in phase retrieval tasks outperforms sparse phase
retrieval methods.
|
Attack and defence in cellular decision-making: lessons from machine
learning
|
Machine learning algorithms can be fooled by small well-designed adversarial
perturbations. This is reminiscent of cellular decision-making where ligands
(called antagonists) prevent correct signalling, like in early immune
recognition. We draw a formal analogy between neural networks used in machine
learning and models of cellular decision-making (adaptive proofreading). We
apply attacks from machine learning to simple decision-making models, and show
explicitly the correspondence to antagonism by weakly bound ligands. Such
antagonism is absent in more nonlinear models, which inspired us to implement a
biomimetic defence in neural networks filtering out adversarial perturbations.
We then apply a gradient-descent approach from machine learning to different
cellular decision-making models, and we reveal the existence of two regimes
characterized by the presence or absence of a critical point for the gradient.
This critical point causes the strongest antagonists to lie close to the
decision boundary. This is validated in the loss landscapes of robust neural
networks and cellular decision-making models, and observed experimentally for
immune cells. For both regimes, we explain how associated defence mechanisms
shape the geometry of the loss landscape, and why different adversarial attacks
are effective in different regimes. Our work connects evolved cellular
decision-making to machine learning, and motivates the design of a general
theory of adversarial perturbations, both for in vivo and in silico systems.
|
A quantum-inspired classical algorithm for recommendation systems
|
We give a classical analogue to Kerenidis and Prakash's quantum
recommendation system, previously believed to be one of the strongest
candidates for provably exponential speedups in quantum machine learning. Our
main result is an algorithm that, given an $m \times n$ matrix in a data
structure supporting certain $\ell^2$-norm sampling operations, outputs an
$\ell^2$-norm sample from a rank-$k$ approximation of that matrix in time
$O(\text{poly}(k)\log(mn))$, only polynomially slower than the quantum
algorithm. As a consequence, Kerenidis and Prakash's algorithm does not in fact
give an exponential speedup over classical algorithms. Further, under strong
input assumptions, the classical recommendation system resulting from our
algorithm produces recommendations exponentially faster than previous classical
systems, which run in time linear in $m$ and $n$.
The main insight of this work is the use of simple routines to manipulate
$\ell^2$-norm sampling distributions, which play the role of quantum
superpositions in the classical setting. This correspondence indicates a
potentially fruitful framework for formally comparing quantum machine learning
algorithms to classical machine learning algorithms.
|
Structured Bayesian Gaussian process latent variable model: applications
to data-driven dimensionality reduction and high-dimensional inversion
|
We introduce a methodology for nonlinear inverse problems using a variational
Bayesian approach where the unknown quantity is a spatial field. A structured
Bayesian Gaussian process latent variable model is used both to construct a
low-dimensional generative model of the sample-based stochastic prior as well
as a surrogate for the forward evaluation. Its Bayesian formulation captures
epistemic uncertainty introduced by the limited number of input and output
examples, automatically selects an appropriate dimensionality for the learned
latent representation of the data, and rigorously propagates the uncertainty of
the data-driven dimensionality reduction of the stochastic space through the
forward model surrogate. The structured Gaussian process model explicitly
leverages spatial information for an informative generative prior to improve
sample efficiency while achieving computational tractability through Kronecker
product decompositions of the relevant kernel matrices. Importantly, the
Bayesian inversion is carried out by solving a variational optimization
problem, replacing traditional computationally-expensive Monte Carlo sampling.
The methodology is demonstrated on an elliptic PDE and is shown to return
well-calibrated posteriors and is tractable with latent spaces with over 100
dimensions.
|
Manifold regularization with GANs for semi-supervised learning
|
Generative Adversarial Networks are powerful generative models that are able
to model the manifold of natural images. We leverage this property to perform
manifold regularization by approximating a variant of the Laplacian norm using
a Monte Carlo approximation that is easily computed with the GAN. When
incorporated into the semi-supervised feature-matching GAN we achieve
state-of-the-art results for GAN-based semi-supervised learning on CIFAR-10 and
SVHN benchmarks, with a method that is significantly easier to implement than
competing methods. We also find that manifold regularization improves the
quality of generated images, and is affected by the quality of the GAN used to
approximate the regularizer.
|
Automated Vulnerability Detection in Source Code Using Deep
Representation Learning
|
Increasing numbers of software vulnerabilities are discovered every year
whether they are reported publicly or discovered internally in proprietary
code. These vulnerabilities can pose serious risk of exploit and result in
system compromise, information leaks, or denial of service. We leveraged the
wealth of C and C++ open-source code available to develop a large-scale
function-level vulnerability detection system using machine learning. To
supplement existing labeled vulnerability datasets, we compiled a vast dataset
of millions of open-source functions and labeled it with carefully-selected
findings from three different static analyzers that indicate potential
exploits. The labeled dataset is available at: https://osf.io/d45bw/. Using
these datasets, we developed a fast and scalable vulnerability detection tool
based on deep feature representation learning that directly interprets lexed
source code. We evaluated our tool on code from both real software packages and
the NIST SATE IV benchmark dataset. Our results demonstrate that deep feature
representation learning on source code is a promising approach for automated
software vulnerability detection.
|
Differentially-Private "Draw and Discard" Machine Learning
|
In this work, we propose a novel framework for privacy-preserving
client-distributed machine learning. It is motivated by the desire to achieve
differential privacy guarantees in the local model of privacy in a way that
satisfies all systems constraints using asynchronous client-server
communication and provides attractive model learning properties. We call it
"Draw and Discard" because it relies on random sampling of models for load
distribution (scalability), which also provides additional server-side privacy
protections and improved model quality through averaging. We present the
mechanics of client and server components of "Draw and Discard" and demonstrate
how the framework can be applied to learning Generalized Linear models. We then
analyze the privacy guarantees provided by our approach against several types
of adversaries and showcase experimental results that provide evidence for the
framework's viability in practical deployments.
|
Topic Diffusion Discovery based on Sparseness-constrained Non-negative
Matrix Factorization
|
Due to recent explosion of text data, researchers have been overwhelmed by
ever-increasing volume of articles produced by different research communities.
Various scholarly search websites, citation recommendation engines, and
research databases have been created to simplify the text search tasks.
However, it is still difficult for researchers to be able to identify potential
research topics without doing intensive reviews on a tremendous number of
articles published by journals, conferences, meetings, and workshops. In this
paper, we consider a novel topic diffusion discovery technique that
incorporates sparseness-constrained Non-negative Matrix Factorization with
generalized Jensen-Shannon divergence to help understand term-topic evolutions
and identify topic diffusions. Our experimental result shows that this approach
can extract more prominent topics from large article databases, visualize
relationships between terms of interest and abstract topics, and further help
researchers understand whether given terms/topics have been widely explored or
whether new topics are emerging from literature.
|
Process Discovery using Classification Tree Hidden Semi-Markov Model
|
Various and ubiquitous information systems are being used in monitoring,
exchanging, and collecting information. These systems are generating massive
amount of event sequence logs that may help us understand underlying
phenomenon. By analyzing these logs, we can learn process models that describe
system procedures, predict the development of the system, or check whether the
changes are expected. In this paper, we consider a novel technique that models
these sequences of events in temporal-probabilistic manners. Specifically, we
propose a probabilistic process model that combines hidden semi-Markov model
and classification trees learning. Our experimental result shows that the
proposed approach can answer a kind of question-"what are the most frequent
sequence of system dynamics relevant to a given sequence of observable
events?". For example, "Given a series of medical treatments, what are the most
relevant patients' health condition pattern changes at different times?"
|
A likelihood-ratio type test for stochastic block models with bounded
degrees
|
A fundamental problem in network data analysis is to test Erd\"{o}s-R\'{e}nyi
model $\mathcal{G}\left(n,\frac{a+b}{2n}\right)$ versus a bisection stochastic
block model $\mathcal{G}\left(n,\frac{a}{n},\frac{b}{n}\right)$, where $a,b>0$
are constants that represent the expected degrees of the graphs and $n$ denotes
the number of nodes. This problem serves as the foundation of many other
problems such as testing-based methods for determining the number of
communities (\cite{BS16,L16}) and community detection (\cite{MS16}). Existing
work has been focusing on growing-degree regime $a,b\to\infty$
(\cite{BS16,L16,MS16,BM17,B18,GL17a,GL17b}) while leaving the bounded-degree
regime untreated. In this paper, we propose a likelihood-ratio (LR) type
procedure based on regularization to test stochastic block models with bounded
degrees. We derive the limit distributions as power Poisson laws under both
null and alternative hypotheses, based on which the limit power of the test is
carefully analyzed. We also examine a Monte-Carlo method that partly resolves
the computational cost issue. The proposed procedures are examined by both
simulated and real-world data. The proof depends on a contiguity theory
developed by Janson \cite{J95}.
|
Simultaneous Coherent Structure Coloring facilitates interpretable
clustering of scientific data by amplifying dissimilarity
|
The clustering of data into physically meaningful subsets often requires
assumptions regarding the number, size, or shape of the subgroups. Here, we
present a new method, simultaneous coherent structure coloring (sCSC), which
accomplishes the task of unsupervised clustering without a priori guidance
regarding the underlying structure of the data. sCSC performs a sequence of
binary splittings on the dataset such that the most dissimilar data points are
required to be in separate clusters. To achieve this, we obtain a set of
orthogonal coordinates along which dissimilarity in the dataset is maximized
from a generalized eigenvalue problem based on the pairwise dissimilarity
between the data points to be clustered. This sequence of bifurcations produces
a binary tree representation of the system, from which the number of clusters
in the data and their interrelationships naturally emerge. To illustrate the
effectiveness of the method in the absence of a priori assumptions, we apply it
to three exemplary problems in fluid dynamics. Then, we illustrate its capacity
for interpretability using a high-dimensional protein folding simulation
dataset. While we restrict our examples to dynamical physical systems in this
work, we anticipate straightforward translation to other fields where existing
analysis tools require ad hoc assumptions on the data structure, lack the
interpretability of the present method, or in which the underlying processes
are less accessible, such as genomics and neuroscience.
|
Convergence Rate of Block-Coordinate Maximization Burer-Monteiro Method
for Solving Large SDPs
|
Semidefinite programming (SDP) with diagonal constraints arise in many
optimization problems, such as Max-Cut, community detection and group
synchronization. Although SDPs can be solved to arbitrary precision in
polynomial time, generic convex solvers do not scale well with the dimension of
the problem. In order to address this issue, Burer and Monteiro proposed to
reduce the dimension of the problem by appealing to a low-rank factorization
and solve the subsequent non-convex problem instead. In this paper, we present
coordinate ascent based methods to solve this non-convex problem with provable
convergence guarantees. More specifically, we prove that the block-coordinate
maximization algorithm applied to the non-convex Burer-Monteiro method globally
converges to a first-order stationary point with a sublinear rate without any
assumptions on the problem. We further show that this algorithm converges
linearly around a local maximum provided that the objective function exhibits
quadratic decay. We establish that this condition generically holds when the
rank of the factorization is sufficiently large. Furthermore, incorporating
Lanczos method to the block-coordinate maximization, we propose an algorithm
that is guaranteed to return a solution that provides $1-O(1/r)$ approximation
to the original SDP without any assumptions, where $r$ is the rank of the
factorization. This approximation ratio is known to be optimal (up to
constants) under the unique games conjecture, and we can explicitly quantify
the number of iterations to obtain such a solution.
|
Will it Blend? Composing Value Functions in Reinforcement Learning
|
An important property for lifelong-learning agents is the ability to combine
existing skills to solve unseen tasks. In general, however, it is unclear how
to compose skills in a principled way. We provide a "recipe" for optimal value
function composition in entropy-regularised reinforcement learning (RL) and
then extend this to the standard RL setting. Composition is demonstrated in a
video game environment, where an agent with an existing library of policies is
able to solve new tasks without the need for further learning.
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.