title
stringlengths 7
246
| abstract
stringlengths 6
3.31k
|
|---|---|
LaVAN: Localized and Visible Adversarial Noise | Most works on adversarial examples for deep-learning based image classifiers
use noise that, while small, covers the entire image. We explore the case where
the noise is allowed to be visible but confined to a small, localized patch of
the image, without covering any of the main object(s) in the image. We show
that it is possible to generate localized adversarial noises that cover only 2%
of the pixels in the image, none of them over the main object, and that are
transferable across images and locations, and successfully fool a
state-of-the-art Inception v3 model with very high success rates.
|
Characterizing Adversarial Subspaces Using Local Intrinsic
Dimensionality | Deep Neural Networks (DNNs) have recently been shown to be vulnerable against
adversarial examples, which are carefully crafted instances that can mislead
DNNs to make errors during prediction. To better understand such attacks, a
characterization is needed of the properties of regions (the so-called
'adversarial subspaces') in which adversarial examples lie. We tackle this
challenge by characterizing the dimensional properties of adversarial regions,
via the use of Local Intrinsic Dimensionality (LID). LID assesses the
space-filling capability of the region surrounding a reference example, based
on the distance distribution of the example to its neighbors. We first provide
explanations about how adversarial perturbation can affect the LID
characteristic of adversarial regions, and then show empirically that LID
characteristics can facilitate the distinction of adversarial examples
generated using state-of-the-art attacks. As a proof-of-concept, we show that a
potential application of LID is to distinguish adversarial examples, and the
preliminary results show that it can outperform several state-of-the-art
detection measures by large margins for five attack strategies considered in
this paper across three benchmark datasets. Our analysis of the LID
characteristic for adversarial regions not only motivates new directions of
effective adversarial defense, but also opens up more challenges for developing
new attacks to better understand the vulnerabilities of DNNs.
|
A Machine Learning Framework for Register Placement Optimization in
Digital Circuit Design | In modern digital circuit back-end design, designers heavily rely on
electronic-design-automoation (EDA) tool to close timing. However, the
heuristic algorithms used in the place and route tool usually does not result
in optimal solution. Thus, significant design effort is used to tune parameters
or provide user constraints or guidelines to improve the tool performance. In
this paper, we targeted at those optimization space left behind by the EDA
tools and propose a machine learning framework that helps to define what are
the guidelines and constraints for registers placement, which can yield better
performance and quality for back-end design. In other words, the framework is
trying to learn what are the flaws of the existing EDA tools and tries to
optimize it by providing additional information. We discuss what is the proper
input feature vector to be extracted, and what is metric to be used for
reference output. We also develop a scheme to generate perturbed training
samples using existing design based on Gaussian randomization. By applying our
methodology, we are able to improve the design runtime by up to 36% and timing
quality by up to 23%.
|
Graph Memory Networks for Molecular Activity Prediction | Molecular activity prediction is critical in drug design. Machine learning
techniques such as kernel methods and random forests have been successful for
this task. These models require fixed-size feature vectors as input while the
molecules are variable in size and structure. As a result, fixed-size
fingerprint representation is poor in handling substructures for large
molecules. In addition, molecular activity tests, or a so-called BioAssays, are
relatively small in the number of tested molecules due to its complexity. Here
we approach the problem through deep neural networks as they are flexible in
modeling structured data such as grids, sequences and graphs. We train multiple
BioAssays using a multi-task learning framework, which combines information
from multiple sources to improve the performance of prediction, especially on
small datasets. We propose Graph Memory Network (GraphMem), a memory-augmented
neural network to model the graph structure in molecules. GraphMem consists of
a recurrent controller coupled with an external memory whose cells dynamically
interact and change through a multi-hop reasoning process. Applied to the
molecules, the dynamic interactions enable an iterative refinement of the
representation of molecular graphs with multiple bond types. GraphMem is
capable of jointly training on multiple datasets by using a specific-task query
fed to the controller as an input. We demonstrate the effectiveness of the
proposed model for separately and jointly training on more than 100K
measurements, spanning across 9 BioAssay activity tests.
|
Boundary Optimizing Network (BON) | Despite all the success that deep neural networks have seen in classifying
certain datasets, the challenge of finding optimal solutions that generalize
still remains. In this paper, we propose the Boundary Optimizing Network (BON),
a new approach to generalization for deep neural networks when used for
supervised learning. Given a classification network, we propose to use a
collaborative generative network that produces new synthetic data points in the
form of perturbations of original data points. In this way, we create a data
support around each original data point which prevents decision boundaries from
passing too close to the original data points, i.e. prevents overfitting. We
show that BON improves convergence on CIFAR-10 using the state-of-the-art
Densenet. We do however observe that the generative network suffers from
catastrophic forgetting during training, and we therefore propose to use a
variation of Memory Aware Synapses to optimize the generative network (called
BON++). On the Iris dataset, we visualize the effect of BON++ when the
generator does not suffer from catastrophic forgetting and conclude that the
approach has the potential to create better boundaries in a higher dimensional
space.
|
Modeling urbanization patterns with generative adversarial networks | In this study we propose a new method to simulate hyper-realistic urban
patterns using Generative Adversarial Networks trained with a global urban
land-use inventory. We generated a synthetic urban "universe" that
qualitatively reproduces the complex spatial organization observed in global
urban patterns, while being able to quantitatively recover certain key
high-level urban spatial metrics.
|
Fusion of ANN and SVM Classifiers for Network Attack Detection | With the progressive increase of network application and electronic devices
(computers, mobile phones, android, etc.) attack and intrusion, detection has
become a very challenging task in cybercrime detection area. in this context,
most of the existing approaches of attack detection rely mainly on a finite set
of attacks. These solutions are vulnerable, that is, they fail in detecting
some attacks when sources of informations are ambiguous or imperfect. However,
few approaches started investigating in this direction. This paper investigates
the role of machine learning approach (ANN, SVM) in detecting a TCP connection
traffic as a normal or a suspicious one. But, using ANN and SVM is an expensive
technique individually. In this paper, combining two classifiers are proposed,
where artificial neural network (ANN) classifier and support vector machine
(SVM) are both employed. Additionally, our proposed solution allows to
visualize obtained classification results. Accuracy of the proposed solution
has been compared with other classifier results. Experiments have been
conducted with different network connections selected from NSL-KDD DARPA
dataset. Empirical results show that combining ANN and SVM techniques for
attack detection is a promising direction.
|
Compressing Deep Neural Networks: A New Hashing Pipeline Using Kac's
Random Walk Matrices | The popularity of deep learning is increasing by the day. However, despite
the recent advancements in hardware, deep neural networks remain
computationally intensive. Recent work has shown that by preserving the angular
distance between vectors, random feature maps are able to reduce dimensionality
without introducing bias to the estimator. We test a variety of established
hashing pipelines as well as a new approach using Kac's random walk matrices.
We demonstrate that this method achieves similar accuracy to existing
pipelines.
|
Rogue Signs: Deceiving Traffic Sign Recognition with Malicious Ads and
Logos | We propose a new real-world attack against the computer vision based systems
of autonomous vehicles (AVs). Our novel Sign Embedding attack exploits the
concept of adversarial examples to modify innocuous signs and advertisements in
the environment such that they are classified as the adversary's desired
traffic sign with high confidence. Our attack greatly expands the scope of the
threat posed to AVs since adversaries are no longer restricted to just
modifying existing traffic signs as in previous work. Our attack pipeline
generates adversarial samples which are robust to the environmental conditions
and noisy image transformations present in the physical world. We ensure this
by including a variety of possible image transformations in the optimization
problem used to generate adversarial samples. We verify the robustness of the
adversarial samples by printing them out and carrying out drive-by tests
simulating the conditions under which image capture would occur in a real-world
scenario. We experimented with physical attack samples for different distances,
lighting conditions and camera angles. In addition, extensive evaluations were
carried out in the virtual setting for a variety of image transformations. The
adversarial samples generated using our method have adversarial success rates
in excess of 95% in the physical as well as virtual settings.
|
Sequential Preference-Based Optimization | Many real-world engineering problems rely on human preferences to guide their
design and optimization. We present PrefOpt, an open source package to simplify
sequential optimization tasks that incorporate human preference feedback. Our
approach extends an existing latent variable model for binary preferences to
allow for observations of equivalent preference from users.
|
Lifelong Learning for Sentiment Classification | This paper proposes a novel lifelong learning (LL) approach to sentiment
classification. LL mimics the human continuous learning process, i.e.,
retaining the knowledge learned from past tasks and use it to help future
learning. In this paper, we first discuss LL in general and then LL for
sentiment classification in particular. The proposed LL approach adopts a
Bayesian optimization framework based on stochastic gradient descent. Our
experimental results show that the proposed method outperforms baseline methods
significantly, which demonstrates that lifelong learning is a promising
research direction.
|
Less is More: Culling the Training Set to Improve Robustness of Deep
Neural Networks | Deep neural networks are vulnerable to adversarial examples. Prior defenses
attempted to make deep networks more robust by either changing the network
architecture or augmenting the training set with adversarial examples, but both
have inherent limitations. Motivated by recent research that shows outliers in
the training set have a high negative influence on the trained model, we
studied the relationship between model robustness and the quality of the
training set. We first show that outliers give the model better generalization
ability but weaker robustness. Next, we propose an adversarial example
detection framework, in which we design two methods for removing outliers from
training set to obtain the sanitized model and then detect adversarial example
by calculating the difference of outputs between the original and the sanitized
model. We evaluated the framework on both MNIST and SVHN. Based on the
difference measured by Kullback-Leibler divergence, we could detect adversarial
examples with accuracy between 94.67% to 99.89%.
|
Convexification of Neural Graph | Traditionally, most complex intelligence architectures are extremely
non-convex, which could not be well performed by convex optimization. However,
this paper decomposes complex structures into three types of nodes: operators,
algorithms and functions. Iteratively, propagating from node to node along
edge, we prove that "regarding the tree-structured neural graph, it is nearly
convex in each variable, when the other variables are fixed." In fact, the
non-convex properties stem from circles and functions, which could be
transformed to be convex with our proposed \textit{\textbf{scale mechanism}}.
Experimentally, we justify our theoretical analysis by two practical
applications.
|
Data Augmentation by Pairing Samples for Images Classification | Data augmentation is a widely used technique in many machine learning tasks,
such as image classification, to virtually enlarge the training dataset size
and avoid overfitting. Traditional data augmentation techniques for image
classification tasks create new samples from the original training data by, for
example, flipping, distorting, adding a small amount of noise to, or cropping a
patch from an original image. In this paper, we introduce a simple but
surprisingly effective data augmentation technique for image classification
tasks. With our technique, named SamplePairing, we synthesize a new sample from
one image by overlaying another image randomly chosen from the training data
(i.e., taking an average of two images for each pixel). By using two images
randomly selected from the training set, we can generate $N^2$ new samples from
$N$ training samples. This simple data augmentation technique significantly
improved classification accuracy for all the tested datasets; for example, the
top-1 error rate was reduced from 33.5% to 29.0% for the ILSVRC 2012 dataset
with GoogLeNet and from 8.22% to 6.93% in the CIFAR-10 dataset. We also show
that our SamplePairing technique largely improved accuracy when the number of
samples in the training set was very small. Therefore, our technique is more
valuable for tasks with a limited amount of training data, such as medical
imaging tasks.
|
Online Cluster Validity Indices for Streaming Data | Cluster analysis is used to explore structure in unlabeled data sets in a
wide range of applications. An important part of cluster analysis is validating
the quality of computationally obtained clusters. A large number of different
internal indices have been developed for validation in the offline setting.
However, this concept has not been extended to the online setting. A key
challenge is to find an efficient incremental formulation of an index that can
capture both cohesion and separation of the clusters over potentially infinite
data streams. In this paper, we develop two online versions (with and without
forgetting factors) of the Xie-Beni and Davies-Bouldin internal validity
indices, and analyze their characteristics, using two streaming clustering
algorithms (sk-means and online ellipsoidal clustering), and illustrate their
use in monitoring evolving clusters in streaming data. We also show that
incremental cluster validity indices are capable of sending a distress signal
to online monitors when evolving clusters go awry. Our numerical examples
indicate that the incremental Xie-Beni index with forgetting factor is superior
to the other three indices tested.
|
An efficient K -means clustering algorithm for massive data | The analysis of continously larger datasets is a task of major importance in
a wide variety of scientific fields. In this sense, cluster analysis algorithms
are a key element of exploratory data analysis, due to their easiness in the
implementation and relatively low computational cost. Among these algorithms,
the K -means algorithm stands out as the most popular approach, besides its
high dependency on the initial conditions, as well as to the fact that it might
not scale well on massive datasets. In this article, we propose a recursive and
parallel approximation to the K -means algorithm that scales well on both the
number of instances and dimensionality of the problem, without affecting the
quality of the approximation. In order to achieve this, instead of analyzing
the entire dataset, we work on small weighted sets of points that mostly intend
to extract information from those regions where it is harder to determine the
correct cluster assignment of the original instances. In addition to different
theoretical properties, which deduce the reasoning behind the algorithm,
experimental results indicate that our method outperforms the state-of-the-art
in terms of the trade-off between number of distance computations and the
quality of the solution obtained.
|
Adversarial Deep Learning for Robust Detection of Binary Encoded Malware | Malware is constantly adapting in order to avoid detection. Model based
malware detectors, such as SVM and neural networks, are vulnerable to so-called
adversarial examples which are modest changes to detectable malware that allows
the resulting malware to evade detection. Continuous-valued methods that are
robust to adversarial examples of images have been developed using saddle-point
optimization formulations. We are inspired by them to develop similar methods
for the discrete, e.g. binary, domain which characterizes the features of
malware. A specific extra challenge of malware is that the adversarial examples
must be generated in a way that preserves their malicious functionality. We
introduce methods capable of generating functionally preserved adversarial
malware examples in the binary domain. Using the saddle-point formulation, we
incorporate the adversarial examples into the training of models that are
robust to them. We evaluate the effectiveness of the methods and others in the
literature on a set of Portable Execution~(PE) files. Comparison prompts our
introduction of an online measure computed during training to assess general
expectation of robustness.
|
Representation Learning with Autoencoders for Electronic Health Records:
A Comparative Study | Increasing volume of Electronic Health Records (EHR) in recent years provides
great opportunities for data scientists to collaborate on different aspects of
healthcare research by applying advanced analytics to these EHR clinical data.
A key requirement however is obtaining meaningful insights from high
dimensional, sparse and complex clinical data. Data science approaches
typically address this challenge by performing feature learning in order to
build more reliable and informative feature representations from clinical data
followed by supervised learning. In this paper, we propose a predictive
modeling approach based on deep learning based feature representations and word
embedding techniques. Our method uses different deep architectures (stacked
sparse autoencoders, deep belief network, adversarial autoencoders and
variational autoencoders) for feature representation in higher-level
abstraction to obtain effective and robust features from EHRs, and then build
prediction models on top of them. Our approach is particularly useful when the
unlabeled data is abundant whereas labeled data is scarce. We investigate the
performance of representation learning through a supervised learning approach.
Our focus is to present a comparative study to evaluate the performance of
different deep architectures through supervised learning and provide insights
in the choice of deep feature representation techniques. Our experiments
demonstrate that for small data sets, stacked sparse autoencoder demonstrates a
superior generality performance in prediction due to sparsity regularization
whereas variational autoencoders outperform the competing approaches for large
data sets due to its capability of learning the representation distribution.
|
How To Make the Gradients Small Stochastically: Even Faster Convex and
Nonconvex SGD | Stochastic gradient descent (SGD) gives an optimal convergence rate when
minimizing convex stochastic objectives $f(x)$. However, in terms of making the
gradients small, the original SGD does not give an optimal rate, even when
$f(x)$ is convex.
If $f(x)$ is convex, to find a point with gradient norm $\varepsilon$, we
design an algorithm SGD3 with a near-optimal rate
$\tilde{O}(\varepsilon^{-2})$, improving the best known rate
$O(\varepsilon^{-8/3})$ of [18].
If $f(x)$ is nonconvex, to find its $\varepsilon$-approximate local minimum,
we design an algorithm SGD5 with rate $\tilde{O}(\varepsilon^{-3.5})$, where
previously SGD variants only achieve $\tilde{O}(\varepsilon^{-4})$ [6, 15, 33].
This is no slower than the best known stochastic version of Newton's method in
all parameter regimes [30].
|
EBIC: an evolutionary-based parallel biclustering algorithm for pattern
discover | In this paper a novel biclustering algorithm based on artificial intelligence
(AI) is introduced. The method called EBIC aims to detect biologically
meaningful, order-preserving patterns in complex data. The proposed algorithm
is probably the first one capable of discovering with accuracy exceeding 50%
multiple complex patterns in real gene expression datasets. It is also one of
the very few biclustering methods designed for parallel environments with
multiple graphics processing units (GPUs). We demonstrate that EBIC outperforms
state-of-the-art biclustering methods, in terms of recovery and relevance, on
both synthetic and genetic datasets. EBIC also yields results over 12 times
faster than the most accurate reference algorithms. The proposed algorithm is
anticipated to be added to the repertoire of unsupervised machine learning
algorithms for the analysis of datasets, including those from large-scale
genomic studies.
|
Meta-Tracker: Fast and Robust Online Adaptation for Visual Object
Trackers | This paper improves state-of-the-art visual object trackers that use online
adaptation. Our core contribution is an offline meta-learning-based method to
adjust the initial deep networks used in online adaptation-based tracking. The
meta learning is driven by the goal of deep networks that can quickly be
adapted to robustly model a particular target in future frames. Ideally the
resulting models focus on features that are useful for future frames, and avoid
overfitting to background clutter, small parts of the target, or noise. By
enforcing a small number of update iterations during meta-learning, the
resulting networks train significantly faster. We demonstrate this approach on
top of the high performance tracking approaches: tracking-by-detection based
MDNet and the correlation based CREST. Experimental results on standard
benchmarks, OTB2015 and VOT2016, show that our meta-learned versions of both
trackers improve speed, accuracy, and robustness.
|
Convergence Analysis of Gradient Descent Algorithms with Proportional
Updates | The rise of deep learning in recent years has brought with it increasingly
clever optimization methods to deal with complex, non-linear loss functions.
These methods are often designed with convex optimization in mind, but have
been shown to work well in practice even for the highly non-convex optimization
associated with neural networks. However, one significant drawback of these
methods when they are applied to deep learning is that the magnitude of the
update step is sometimes disproportionate to the magnitude of the weights (much
smaller or larger), leading to training instabilities such as vanishing and
exploding gradients. An idea to combat this issue is gradient descent with
proportional updates. Gradient descent with proportional updates was introduced
in 2017. It was independently developed by You et al (Layer-wise Adaptive Rate
Scaling (LARS) algorithm) and by Abu-El-Haija (PercentDelta algorithm). The
basic idea of both of these algorithms is to make each step of the gradient
descent proportional to the current weight norm and independent of the gradient
magnitude. It is common in the context of new optimization methods to prove
convergence or derive regret bounds under the assumption of Lipschitz
continuity and convexity. However, even though LARS and PercentDelta were shown
to work well in practice, there is no theoretical analysis of the convergence
properties of these algorithms. Thus it is not clear if the idea of gradient
descent with proportional updates is used in the optimal way, or if it could be
improved by using a different norm or specific learning rate schedule, for
example. Moreover, it is not clear if these algorithms can be extended to other
problems, besides neural networks. We attempt to answer these questions by
establishing the theoretical analysis of gradient descent with proportional
updates, and verifying this analysis with empirical examples.
|
Comparing heterogeneous entities using artificial neural networks of
trainable weighted structural components and machine-learned activation
functions | To compare entities of differing types and structural components, the
artificial neural network paradigm was used to cross-compare structural
components between heterogeneous documents. Trainable weighted structural
components were input into machine-learned activation functions of the neurons.
The model was used for matching news articles and videos, where the inputs and
activation functions respectively consisted of term vectors and cosine
similarity measures between the weighted structural components. The model was
tested with different weights, achieving as high as 59.2% accuracy for matching
videos to news articles. A mobile application user interface for recommending
related videos for news articles was developed to demonstrate consumer value,
including its potential usefulness for cross-selling products from unrelated
categories.
|
Paranom: A Parallel Anomaly Dataset Generator | In this paper, we present Paranom, a parallel anomaly dataset generator. We
discuss its design and provide brief experimental results demonstrating its
usefulness in improving the classification correctness of LSTM-AD, a
state-of-the-art anomaly detection model.
|
Adaptive Graph Convolutional Neural Networks | Graph Convolutional Neural Networks (Graph CNNs) are generalizations of
classical CNNs to handle graph data such as molecular data, point could and
social networks. Current filters in graph CNNs are built for fixed and shared
graph structure. However, for most real data, the graph structures varies in
both size and connectivity. The paper proposes a generalized and flexible graph
CNN taking data of arbitrary graph structure as input. In that way a
task-driven adaptive graph is learned for each graph data while training. To
efficiently learn the graph, a distance metric learning is proposed. Extensive
experiments on nine graph-structured datasets have demonstrated the superior
performance improvement on both convergence speed and predictive accuracy.
|
Lung and Pancreatic Tumor Characterization in the Deep Learning Era:
Novel Supervised and Unsupervised Learning Approaches | Risk stratification (characterization) of tumors from radiology images can be
more accurate and faster with computer-aided diagnosis (CAD) tools. Tumor
characterization through such tools can also enable non-invasive cancer
staging, prognosis, and foster personalized treatment planning as a part of
precision medicine. In this study, we propose both supervised and unsupervised
machine learning strategies to improve tumor characterization. Our first
approach is based on supervised learning for which we demonstrate significant
gains with deep learning algorithms, particularly by utilizing a 3D
Convolutional Neural Network and Transfer Learning. Motivated by the
radiologists' interpretations of the scans, we then show how to incorporate
task dependent feature representations into a CAD system via a
graph-regularized sparse Multi-Task Learning (MTL) framework. In the second
approach, we explore an unsupervised learning algorithm to address the limited
availability of labeled training data, a common problem in medical imaging
applications. Inspired by learning from label proportion (LLP) approaches in
computer vision, we propose to use proportion-SVM for characterizing tumors. We
also seek the answer to the fundamental question about the goodness of "deep
features" for unsupervised tumor classification. We evaluate our proposed
supervised and unsupervised learning algorithms on two different tumor
diagnosis challenges: lung and pancreas with 1018 CT and 171 MRI scans,
respectively, and obtain the state-of-the-art sensitivity and specificity
results in both problems.
|
eCommerceGAN : A Generative Adversarial Network for E-commerce | E-commerce companies such as Amazon, Alibaba and Flipkart process billions of
orders every year. However, these orders represent only a small fraction of all
plausible orders. Exploring the space of all plausible orders could help us
better understand the relationships between the various entities in an
e-commerce ecosystem, namely the customers and the products they purchase. In
this paper, we propose a Generative Adversarial Network (GAN) for orders made
in e-commerce websites. Once trained, the generator in the GAN could generate
any number of plausible orders. Our contributions include: (a) creating a dense
and low-dimensional representation of e-commerce orders, (b) train an
ecommerceGAN (ecGAN) with real orders to show the feasibility of the proposed
paradigm, and (c) train an ecommerce-conditional-GAN (ec^2GAN) to generate the
plausible orders involving a particular product. We propose several qualitative
methods to evaluate ecGAN and demonstrate its effectiveness. The ec^2GAN is
used for various kinds of characterization of possible orders involving a
product that has just been introduced into the e-commerce system. The proposed
approach ec^2GAN performs significantly better than the baseline in most of the
scenarios.
|
More Adaptive Algorithms for Adversarial Bandits | We develop a novel and generic algorithm for the adversarial multi-armed
bandit problem (or more generally the combinatorial semi-bandit problem). When
instantiated differently, our algorithm achieves various new data-dependent
regret bounds improving previous work. Examples include: 1) a regret bound
depending on the variance of only the best arm; 2) a regret bound depending on
the first-order path-length of only the best arm; 3) a regret bound depending
on the sum of first-order path-lengths of all arms as well as an important
negative term, which together lead to faster convergence rates for some normal
form games with partial feedback; 4) a regret bound that simultaneously implies
small regret when the best arm has small loss and logarithmic regret when there
exists an arm whose expected loss is always smaller than those of others by a
fixed gap (e.g. the classic i.i.d. setting). In some cases, such as the last
two results, our algorithm is completely parameter-free.
The main idea of our algorithm is to apply the optimism and adaptivity
techniques to the well-known Online Mirror Descent framework with a special
log-barrier regularizer. The challenges are to come up with appropriate
optimistic predictions and correction terms in this framework. Some of our
results also crucially rely on using a sophisticated increasing learning rate
schedule.
|
Weakly Supervised One-Shot Detection with Attention Similarity Networks | Neural network models that are not conditioned on class identities were shown
to facilitate knowledge transfer between classes and to be well-suited for
one-shot learning tasks. Following this motivation, we further explore and
establish such models and present a novel neural network architecture for the
task of weakly supervised one-shot detection. Our model is only conditioned on
a single exemplar of an unseen class and a larger target example that may or
may not contain an instance of the same class as the exemplar. By pairing a
Siamese similarity network with an attention mechanism, we design a model that
manages to simultaneously identify and localise instances of classes unseen at
training time. In experiments with datasets from the computer vision and audio
domains, the proposed method considerably outperforms the baseline methods for
the weakly supervised one-shot detection task.
|
Fooling End-to-end Speaker Verification by Adversarial Examples | Automatic speaker verification systems are increasingly used as the primary
means to authenticate costumers. Recently, it has been proposed to train
speaker verification systems using end-to-end deep neural models. In this
paper, we show that such systems are vulnerable to adversarial example attack.
Adversarial examples are generated by adding a peculiar noise to original
speaker examples, in such a way that they are almost indistinguishable from the
original examples by a human listener. Yet, the generated waveforms, which
sound as speaker A can be used to fool such a system by claiming as if the
waveforms were uttered by speaker B. We present white-box attacks on an
end-to-end deep network that was either trained on YOHO or NTIMIT. We also
present two black-box attacks: where the adversarial examples were generated
with a system that was trained on YOHO, but the attack is on a system that was
trained on NTIMIT; and when the adversarial examples were generated with a
system that was trained on Mel-spectrum feature set, but the attack is on a
system that was trained on MFCC. Results suggest that the accuracy of the
attacked system was decreased and the false-positive rate was dramatically
increased.
|
Blessing of dimensionality: mathematical foundations of the statistical
physics of data | The concentration of measure phenomena were discovered as the mathematical
background of statistical mechanics at the end of the XIX - beginning of the XX
century and were then explored in mathematics of the XX-XXI centuries. At the
beginning of the XXI century, it became clear that the proper utilisation of
these phenomena in machine learning might transform the curse of dimensionality
into the blessing of dimensionality.
This paper summarises recently discovered phenomena of measure concentration
which drastically simplify some machine learning problems in high dimension,
and allow us to correct legacy artificial intelligence systems. The classical
concentration of measure theorems state that i.i.d. random points are
concentrated in a thin layer near a surface (a sphere or equators of a sphere,
an average or median level set of energy or another Lipschitz function, etc.).
The new stochastic separation theorems describe the thin structure of these
thin layers: the random points are not only concentrated in a thin layer but
are all linearly separable from the rest of the set, even for exponentially
large random sets. The linear functionals for separation of points can be
selected in the form of the linear Fisher's discriminant.
All artificial intelligence systems make errors. Non-destructive correction
requires separation of the situations (samples) with errors from the samples
corresponding to correct behaviour by a simple and robust classifier. The
stochastic separation theorems provide us by such classifiers and a
non-iterative (one-shot) procedure for learning.
|
A Smoothed Analysis of the Greedy Algorithm for the Linear Contextual
Bandit Problem | Bandit learning is characterized by the tension between long-term exploration
and short-term exploitation. However, as has recently been noted, in settings
in which the choices of the learning algorithm correspond to important
decisions about individual people (such as criminal recidivism prediction,
lending, and sequential drug trials), exploration corresponds to explicitly
sacrificing the well-being of one individual for the potential future benefit
of others. This raises a fairness concern. In such settings, one might like to
run a "greedy" algorithm, which always makes the (myopically) optimal decision
for the individuals at hand - but doing this can result in a catastrophic
failure to learn. In this paper, we consider the linear contextual bandit
problem and revisit the performance of the greedy algorithm. We give a smoothed
analysis, showing that even when contexts may be chosen by an adversary, small
perturbations of the adversary's choices suffice for the algorithm to achieve
"no regret", perhaps (depending on the specifics of the setting) with a
constant amount of initial training data. This suggests that "generically"
(i.e. in slightly perturbed environments), exploration and exploitation need
not be in conflict in the linear setting.
|
Approximation beats concentration? An approximation view on inference
with smooth radial kernels | Positive definite kernels and their associated Reproducing Kernel Hilbert
Spaces provide a mathematically compelling and practically competitive
framework for learning from data.
In this paper we take the approximation theory point of view to explore
various aspects of smooth kernels related to their inferential properties. We
analyze eigenvalue decay of kernels operators and matrices, properties of
eigenfunctions/eigenvectors and "Fourier" coefficients of functions in the
kernel space restricted to a discrete set of data points. We also investigate
the fitting capacity of kernels, giving explicit bounds on the fat shattering
dimension of the balls in Reproducing Kernel Hilbert spaces. Interestingly, the
same properties that make kernels very effective approximators for functions in
their "native" kernel space, also limit their capacity to represent arbitrary
functions. We discuss various implications, including those for gradient
descent type methods.
It is important to note that most of our bounds are measure independent.
Moreover, at least in moderate dimension, the bounds for eigenvalues are much
tighter than the bounds which can be obtained from the usual matrix
concentration results. For example, we see that the eigenvalues of kernel
matrices show nearly exponential decay with constants depending only on the
kernel and the domain. We call this "approximation beats concentration"
phenomenon as even when the data are sampled from a probability distribution,
some of their aspects are better understood in terms of approximation theory.
|
Selection Problems in the Presence of Implicit Bias | Over the past two decades, the notion of implicit bias has come to serve as
an important component in our understanding of discrimination in activities
such as hiring, promotion, and school admissions. Research on implicit bias
posits that when people evaluate others -- for example, in a hiring context --
their unconscious biases about membership in particular groups can have an
effect on their decision-making, even when they have no deliberate intention to
discriminate against members of these groups. A growing body of experimental
work has pointed to the effect that implicit bias can have in producing adverse
outcomes.
Here we propose a theoretical model for studying the effects of implicit bias
on selection decisions, and a way of analyzing possible procedural remedies for
implicit bias within this model. A canonical situation represented by our model
is a hiring setting: a recruiting committee is trying to choose a set of
finalists to interview among the applicants for a job, evaluating these
applicants based on their future potential, but their estimates of potential
are skewed by implicit bias against members of one group. In this model, we
show that measures such as the Rooney Rule, a requirement that at least one of
the finalists be chosen from the affected group, can not only improve the
representation of this affected group, but also lead to higher payoffs in
absolute terms for the organization performing the recruiting. However,
identifying the conditions under which such measures can lead to improved
payoffs involves subtle trade-offs between the extent of the bias and the
underlying distribution of applicant characteristics, leading to novel
theoretical questions about order statistics in the presence of probabilistic
side information.
|
Inference Suboptimality in Variational Autoencoders | Amortized inference allows latent-variable models trained via variational
learning to scale to large datasets. The quality of approximate inference is
determined by two factors: a) the capacity of the variational distribution to
match the true posterior and b) the ability of the recognition network to
produce good variational parameters for each datapoint. We examine approximate
inference in variational autoencoders in terms of these factors. We find that
divergence from the true posterior is often due to imperfect recognition
networks, rather than the limited complexity of the approximating distribution.
We show that this is due partly to the generator learning to accommodate the
choice of approximation. Furthermore, we show that the parameters used to
increase the expressiveness of the approximation play a role in generalizing
inference rather than simply improving the complexity of the approximation.
|
Deep Classification of Epileptic Signals | Electrophysiological observation plays a major role in epilepsy evaluation.
However, human interpretation of brain signals is subjective and prone to
misdiagnosis. Automating this process, especially seizure detection relying on
scalp-based Electroencephalography (EEG) and intracranial EEG, has been the
focus of research over recent decades. Nevertheless, its numerous challenges
have inhibited a definitive solution. Inspired by recent advances in deep
learning, we propose a new classification approach for EEG time series based on
Recurrent Neural Networks (RNNs) via the use of Long-Short Term Memory (LSTM)
networks. The proposed deep network effectively learns and models
discriminative temporal patterns from EEG sequential data. Especially, the
features are automatically discovered from the raw EEG data without any
pre-processing step, eliminating humans from laborious feature design task. We
also show that, in the epilepsy scenario, simple architectures can achieve
competitive performance. Using simple architectures significantly benefits in
the practical scenario considering their low computation complexity and reduced
requirement for large training datasets. Using a public dataset, a multi-fold
cross-validation scheme exhibited an average validation accuracy of 95.54\% and
an average AUC of 0.9582 of the ROC curve among all sets defined in the
experiment. This work reinforces the benefits of deep learning to be further
attended in clinical applications and neuroscientific research.
|
Which Neural Net Architectures Give Rise To Exploding and Vanishing
Gradients? | We give a rigorous analysis of the statistical behavior of gradients in a
randomly initialized fully connected network N with ReLU activations. Our
results show that the empirical variance of the squares of the entries in the
input-output Jacobian of N is exponential in a simple architecture-dependent
constant beta, given by the sum of the reciprocals of the hidden layer widths.
When beta is large, the gradients computed by N at initialization vary wildly.
Our approach complements the mean field theory analysis of random networks.
From this point of view, we rigorously compute finite width corrections to the
statistics of gradients at the edge of chaos.
|
Improved asynchronous parallel optimization analysis for stochastic
incremental methods | As datasets continue to increase in size and multi-core computer
architectures are developed, asynchronous parallel optimization algorithms
become more and more essential to the field of Machine Learning. Unfortunately,
conducting the theoretical analysis asynchronous methods is difficult, notably
due to the introduction of delay and inconsistency in inherently sequential
algorithms. Handling these issues often requires resorting to simplifying but
unrealistic assumptions. Through a novel perspective, we revisit and clarify a
subtle but important technical issue present in a large fraction of the recent
convergence rate proofs for asynchronous parallel optimization algorithms, and
propose a simplification of the recently introduced "perturbed iterate"
framework that resolves it. We demonstrate the usefulness of our new framework
by analyzing three distinct asynchronous parallel incremental optimization
algorithms: Hogwild (asynchronous SGD), KROMAGNON (asynchronous SVRG) and
ASAGA, a novel asynchronous parallel version of the incremental gradient
algorithm SAGA that enjoys fast linear convergence rates. We are able to both
remove problematic assumptions and obtain better theoretical results. Notably,
we prove that ASAGA and KROMAGNON can obtain a theoretical linear speedup on
multi-core systems even without sparsity assumptions. We present results of an
implementation on a 40-core architecture illustrating the practical speedups as
well as the hardware overhead. Finally, we investigate the overlap constant, an
ill-understood but central quantity for the theoretical analysis of
asynchronous parallel algorithms. We find that it encompasses much more
complexity than suggested in previous work, and often is order-of-magnitude
bigger than traditionally thought.
|
Autoencoders and Probabilistic Inference with Missing Data: An Exact
Solution for The Factor Analysis Case | Latent variable models can be used to probabilistically "fill-in" missing
data entries. The variational autoencoder architecture (Kingma and Welling,
2014; Rezende et al., 2014) includes a "recognition" or "encoder" network that
infers the latent variables given the data variables. However, it is not clear
how to handle missing data variables in this network. The factor analysis (FA)
model is a basic autoencoder, using linear encoder and decoder networks. We
show how to calculate exactly the latent posterior distribution for the factor
analysis (FA) model in the presence of missing data, and note that this
solution implies that a different encoder network is required for each pattern
of missingness. We also discuss various approximations to the exact solution.
Experiments compare the effectiveness of various approaches to filling in the
missing data.
|
MXNET-MPI: Embedding MPI parallelism in Parameter Server Task Model for
scaling Deep Learning | Existing Deep Learning frameworks exclusively use either Parameter Server(PS)
approach or MPI parallelism. In this paper, we discuss the drawbacks of such
approaches and propose a generic framework supporting both PS and MPI
programming paradigms, co-existing at the same time. The key advantage of the
new model is to embed the scaling benefits of MPI parallelism into the loosely
coupled PS task model. Apart from providing a practical usage model of MPI in
cloud, such framework allows for novel communication avoiding algorithms that
do parameter averaging in Stochastic Gradient Descent(SGD) approaches. We show
how MPI and PS models can synergestically apply algorithms such as Elastic SGD
to improve the rate of convergence against existing approaches. These new
algorithms directly help scaling SGD clusterwide. Further, we also optimize the
critical component of the framework, namely global aggregation or allreduce
using a novel concept of tensor collectives. These treat a group of vectors on
a node as a single object allowing for the existing single vector algorithms to
be directly applicable. We back our claims with sufficient emperical evidence
using large scale ImageNet 1K data. Our framework is built upon MXNET but the
design is generic and can be adapted to other popular DL infrastructures.
|
Learning and Inferring a Driver's Braking Action in Car-Following
Scenarios | Accurately predicting and inferring a driver's decision to brake is critical
for designing warning systems and avoiding collisions. In this paper we focus
on predicting a driver's intent to brake in car-following scenarios from a
perception-decision-action perspective according to his/her driving history. A
learning-based inference method, using onboard data from CAN-Bus, radar and
cameras as explanatory variables, is introduced to infer drivers' braking
decisions by combining a Gaussian mixture model (GMM) with a hidden Markov
model (HMM). The GMM is used to model stochastic relationships among variables,
while the HMM is applied to infer drivers' braking actions based on the GMM.
Real-case driving data from 49 drivers (more than three years' driving data per
driver on average) have been collected from the University of Michigan Safety
Pilot Model Deployment database. We compare the GMM-HMM method to a support
vector machine (SVM) method and an SVM-Bayesian filtering method. The
experimental results are evaluated by employing three performance metrics:
accuracy, sensitivity, specificity. The comparison results show that the
GMM-HMM obtains the best performance, with an accuracy of 90%, sensitivity of
84%, and specificity of 97%. Thus, we believe that this method has great
potential for real-world active safety systems.
|
Stochastic Learning of Nonstationary Kernels for Natural Language
Modeling | Natural language processing often involves computations with semantic or
syntactic graphs to facilitate sophisticated reasoning based on structural
relationships. While convolution kernels provide a powerful tool for comparing
graph structure based on node (word) level relationships, they are difficult to
customize and can be computationally expensive. We propose a generalization of
convolution kernels, with a nonstationary model, for better expressibility of
natural languages in supervised settings. For a scalable learning of the
parameters introduced with our model, we propose a novel algorithm that
leverages stochastic sampling on k-nearest neighbor graphs, along with
approximations based on locality-sensitive hashing. We demonstrate the
advantages of our approach on a challenging real-world (structured inference)
problem of automatically extracting biological models from the text of
scientific papers.
|
The Complexity of Learning Acyclic Conditional Preference Networks | Learning of user preferences, as represented by, for example, Conditional
Preference Networks (CP-nets), has become a core issue in AI research. Recent
studies investigate learning of CP-nets from randomly chosen examples or from
membership and equivalence queries. To assess the optimality of learning
algorithms as well as to better understand the combinatorial structure of
classes of CP-nets, it is helpful to calculate certain learning-theoretic
information complexity parameters. This article focuses on the frequently
studied case of learning from so-called swap examples, which express
preferences among objects that differ in only one attribute. It presents bounds
on or exact values of some well-studied information complexity parameters,
namely the VC dimension, the teaching dimension, and the recursive teaching
dimension, for classes of acyclic CP-nets. We further provide algorithms that
learn tree-structured and general acyclic CP-nets from membership queries.
Using our results on complexity parameters, we assess the optimality of our
algorithms as well as that of another query learning algorithm for acyclic
CP-nets presented in the literature. Our algorithms are near-optimal, and can,
under certain assumptions, be adapted to the case when the membership oracle is
faulty.
|
Some techniques in density estimation | Density estimation is an interdisciplinary topic at the intersection of
statistics, theoretical computer science and machine learning. We review some
old and new techniques for bounding the sample complexity of estimating
densities of continuous distributions, focusing on the class of mixtures of
Gaussians and its subclasses. In particular, we review the main techniques used
to prove the new sample complexity bounds for mixtures of Gaussians by
Ashtiani, Ben-David, Harvey, Liaw, Mehrabian, and Plan arXiv:1710.05209.
|
A Hardware-Friendly Algorithm for Scalable Training and Deployment of
Dimensionality Reduction Models on FPGA | With ever-increasing application of machine learning models in various
domains such as image classification, speech recognition and synthesis, and
health care, designing efficient hardware for these models has gained a lot of
popularity. While the majority of researches in this area focus on efficient
deployment of machine learning models (a.k.a inference), this work concentrates
on challenges of training these models in hardware. In particular, this paper
presents a high-performance, scalable, reconfigurable solution for both
training and deployment of different dimensionality reduction models in
hardware by introducing a hardware-friendly algorithm. Compared to
state-of-the-art implementations, our proposed algorithm and its hardware
realization decrease resource consumption by 50\% without any degradation in
accuracy.
|
Theoretical Impediments to Machine Learning With Seven Sparks from the
Causal Revolution | Current machine learning systems operate, almost exclusively, in a
statistical, or model-free mode, which entails severe theoretical limits on
their power and performance. Such systems cannot reason about interventions and
retrospection and, therefore, cannot serve as the basis for strong AI. To
achieve human level intelligence, learning machines need the guidance of a
model of reality, similar to the ones used in causal inference tasks. To
demonstrate the essential role of such models, I will present a summary of
seven tasks which are beyond reach of current machine learning systems and
which have been accomplished using the tools of causal modeling.
|
Noisy Expectation-Maximization: Applications and Generalizations | We present a noise-injected version of the Expectation-Maximization (EM)
algorithm: the Noisy Expectation Maximization (NEM) algorithm. The NEM
algorithm uses noise to speed up the convergence of the EM algorithm. The NEM
theorem shows that injected noise speeds up the average convergence of the EM
algorithm to a local maximum of the likelihood surface if a positivity
condition holds. The generalized form of the noisy expectation-maximization
(NEM) algorithm allow for arbitrary modes of noise injection including adding
and multiplying noise to the data.
We demonstrate these noise benefits on EM algorithms for the Gaussian mixture
model (GMM) with both additive and multiplicative NEM noise injection. A
separate theorem (not presented here) shows that the noise benefit for
independent identically distributed additive noise decreases with sample size
in mixture models. This theorem implies that the noise benefit is most
pronounced if the data is sparse. Injecting blind noise only slowed
convergence.
|
A3T: Adversarially Augmented Adversarial Training | Recent research showed that deep neural networks are highly sensitive to
so-called adversarial perturbations, which are tiny perturbations of the input
data purposely designed to fool a machine learning classifier. Most
classification models, including deep learning models, are highly vulnerable to
adversarial attacks. In this work, we investigate a procedure to improve
adversarial robustness of deep neural networks through enforcing representation
invariance. The idea is to train the classifier jointly with a discriminator
attached to one of its hidden layer and trained to filter the adversarial
noise. We perform preliminary experiments to test the viability of the approach
and to compare it to other standard adversarial training methods.
|
MINE: Mutual Information Neural Estimation | We argue that the estimation of mutual information between high dimensional
continuous random variables can be achieved by gradient descent over neural
networks. We present a Mutual Information Neural Estimator (MINE) that is
linearly scalable in dimensionality as well as in sample size, trainable
through back-prop, and strongly consistent. We present a handful of
applications on which MINE can be used to minimize or maximize mutual
information. We apply MINE to improve adversarially trained generative models.
We also use MINE to implement Information Bottleneck, applying it to supervised
classification; our results demonstrate substantial improvement in flexibility
and performance in these settings.
|
Arhuaco: Deep Learning and Isolation Based Security for Distributed
High-Throughput Computing | Grid computing systems require innovative methods and tools to identify
cybersecurity incidents and perform autonomous actions i.e. without
administrator intervention. They also require methods to isolate and trace job
payload activity in order to protect users and find evidence of malicious
behavior. We introduce an integrated approach of security monitoring via
Security by Isolation with Linux Containers and Deep Learning methods for the
analysis of real time data in Grid jobs running inside virtualized
High-Throughput Computing infrastructure in order to detect and prevent
intrusions. A dataset for malware detection in Grid computing is described. We
show in addition the utilization of generative methods with Recurrent Neural
Networks to improve the collected dataset. We present Arhuaco, a prototype
implementation of the proposed methods. We empirically study the performance of
our technique. The results show that Arhuaco outperforms other methods used in
Intrusion Detection Systems for Grid Computing. The study is carried out in the
ALICE Collaboration Grid, part of the Worldwide LHC Computing Grid.
|
Towards Arbitrary Noise Augmentation - Deep Learning for Sampling from
Arbitrary Probability Distributions | Accurate noise modelling is important for training of deep learning
reconstruction algorithms. While noise models are well known for traditional
imaging techniques, the noise distribution of a novel sensor may be difficult
to determine a priori. Therefore, we propose learning arbitrary noise
distributions. To do so, this paper proposes a fully connected neural network
model to map samples from a uniform distribution to samples of any explicitly
known probability density function. During the training, the Jensen-Shannon
divergence between the distribution of the model's output and the target
distribution is minimized. We experimentally demonstrate that our model
converges towards the desired state. It provides an alternative to existing
sampling methods such as inversion sampling, rejection sampling, Gaussian
mixture models and Markov-Chain-Monte-Carlo. Our model has high sampling
efficiency and is easily applied to any probability distribution, without the
need of further analytical or numerical calculations.
|
Conditional Probability Models for Deep Image Compression | Deep Neural Networks trained as image auto-encoders have recently emerged as
a promising direction for advancing the state-of-the-art in image compression.
The key challenge in learning such networks is twofold: To deal with
quantization, and to control the trade-off between reconstruction error
(distortion) and entropy (rate) of the latent image representation. In this
paper, we focus on the latter challenge and propose a new technique to navigate
the rate-distortion trade-off for an image compression auto-encoder. The main
idea is to directly model the entropy of the latent representation by using a
context model: A 3D-CNN which learns a conditional probability model of the
latent distribution of the auto-encoder. During training, the auto-encoder
makes use of the context model to estimate the entropy of its representation,
and the context model is concurrently updated to learn the dependencies between
the symbols in the latent representation. Our experiments show that this
approach, when measured in MS-SSIM, yields a state-of-the-art image compression
system based on a simple convolutional auto-encoder.
|
Comparative Study on Generative Adversarial Networks | In recent years, there have been tremendous advancements in the field of
machine learning. These advancements have been made through both academic as
well as industrial research. Lately, a fair amount of research has been
dedicated to the usage of generative models in the field of computer vision and
image classification. These generative models have been popularized through a
new framework called Generative Adversarial Networks. Moreover, many modified
versions of this framework have been proposed in the last two years. We study
the original model proposed by Goodfellow et al. as well as modifications over
the original model and provide a comparative analysis of these models.
|
Asynchronous Stochastic Variational Inference | Stochastic variational inference (SVI) employs stochastic optimization to
scale up Bayesian computation to massive data. Since SVI is at its core a
stochastic gradient-based algorithm, horizontal parallelism can be harnessed to
allow larger scale inference. We propose a lock-free parallel implementation
for SVI which allows distributed computations over multiple slaves in an
asynchronous style. We show that our implementation leads to linear speed-up
while guaranteeing an asymptotic ergodic convergence rate $O(1/\sqrt(T)$ )
given that the number of slaves is bounded by $\sqrt(T)$ ($T$ is the total
number of iterations). The implementation is done in a high-performance
computing (HPC) environment using message passing interface (MPI) for python
(MPI4py). The extensive empirical evaluation shows that our parallel SVI is
lossless, performing comparably well to its counterpart serial SVI with linear
speed-up.
|
Generalization Error Bounds for Noisy, Iterative Algorithms | In statistical learning theory, generalization error is used to quantify the
degree to which a supervised machine learning algorithm may overfit to training
data. Recent work [Xu and Raginsky (2017)] has established a bound on the
generalization error of empirical risk minimization based on the mutual
information $I(S;W)$ between the algorithm input $S$ and the algorithm output
$W$, when the loss function is sub-Gaussian. We leverage these results to
derive generalization error bounds for a broad class of iterative algorithms
that are characterized by bounded, noisy updates with Markovian structure. Our
bounds are very general and are applicable to numerous settings of interest,
including stochastic gradient Langevin dynamics (SGLD) and variants of the
stochastic gradient Hamiltonian Monte Carlo (SGHMC) algorithm. Furthermore, our
error bounds hold for any output function computed over the path of iterates,
including the last iterate of the algorithm or the average of subsets of
iterates, and also allow for non-uniform sampling of data in successive updates
of the algorithm.
|
Not All Ops Are Created Equal! | Efficient and compact neural network models are essential for enabling the
deployment on mobile and embedded devices. In this work, we point out that
typical design metrics for gauging the efficiency of neural network
architectures -- total number of operations and parameters -- are not
sufficient. These metrics may not accurately correlate with the actual
deployment metrics such as energy and memory footprint. We show that throughput
and energy varies by up to 5X across different neural network operation types
on an off-the-shelf Arm Cortex-M7 microcontroller. Furthermore, we show that
the memory required for activation data also need to be considered, apart from
the model parameters, for network architecture exploration studies.
|
Estimating the Number of Connected Components in a Graph via Subgraph
Sampling | Learning properties of large graphs from samples has been an important
problem in statistical network analysis since the early work of Goodman
\cite{Goodman1949} and Frank \cite{Frank1978}. We revisit a problem formulated
by Frank \cite{Frank1978} of estimating the number of connected components in a
large graph based on the subgraph sampling model, in which we randomly sample a
subset of the vertices and observe the induced subgraph. The key question is
whether accurate estimation is achievable in the \emph{sublinear} regime where
only a vanishing fraction of the vertices are sampled. We show that it is
impossible if the parent graph is allowed to contain high-degree vertices or
long induced cycles. For the class of chordal graphs, where induced cycles of
length four or above are forbidden, we characterize the optimal sample
complexity within constant factors and construct linear-time estimators that
provably achieve these bounds. This significantly expands the scope of previous
results which have focused on unbiased estimators and special classes of graphs
such as forests or cliques.
Both the construction and the analysis of the proposed methodology rely on
combinatorial properties of chordal graphs and identities of induced subgraph
counts. They, in turn, also play a key role in proving minimax lower bounds
based on construction of random instances of graphs with matching structures of
small subgraphs.
|
Predicting Future Lane Changes of Other Highway Vehicles using RNN-based
Deep Models | In the event of sensor failure, autonomous vehicles need to safely execute
emergency maneuvers while avoiding other vehicles on the road. To accomplish
this, the sensor-failed vehicle must predict the future semantic behaviors of
other drivers, such as lane changes, as well as their future trajectories given
a recent window of past sensor observations. We address the first issue of
semantic behavior prediction in this paper, which is a precursor to trajectory
prediction, by introducing a framework that leverages the power of recurrent
neural networks (RNNs) and graphical models. Our goal is to predict the future
categorical driving intent, for lane changes, of neighboring vehicles up to
three seconds into the future given as little as a one-second window of past
LIDAR, GPS, inertial, and map data.
We collect real-world data containing over 20 hours of highway driving using
an autonomous Toyota vehicle. We propose a composite RNN model by adopting the
methodology of Structural Recurrent Neural Networks (RNNs) to learn factor
functions and take advantage of both the high-level structure of graphical
models and the sequence modeling power of RNNs, which we expect to afford more
transparent modeling and activity than opaque, single RNN models. To
demonstrate our approach, we validate our model using authentic interstate
highway driving to predict the future lane change maneuvers of other vehicles
neighboring our autonomous vehicle. We find that our composite Structural RNN
outperforms baselines by as much as 12% in balanced accuracy metrics.
|
Combining Symbolic Expressions and Black-box Function Evaluations in
Neural Programs | Neural programming involves training neural networks to learn programs,
mathematics, or logic from data. Previous works have failed to achieve good
generalization performance, especially on problems and programs with high
complexity or on large domains. This is because they mostly rely either on
black-box function evaluations that do not capture the structure of the
program, or on detailed execution traces that are expensive to obtain, and
hence the training data has poor coverage of the domain under consideration. We
present a novel framework that utilizes black-box function evaluations, in
conjunction with symbolic expressions that define relationships between the
given functions. We employ tree LSTMs to incorporate the structure of the
symbolic expression trees. We use tree encoding for numbers present in function
evaluation data, based on their decimal representation. We present an
evaluation benchmark for this task to demonstrate our proposed model combines
symbolic reasoning and function evaluation in a fruitful manner, obtaining high
accuracies in our experiments. Our framework generalizes significantly better
to expressions of higher depth and is able to fill partial equations with valid
completions.
|
Black-box Generation of Adversarial Text Sequences to Evade Deep
Learning Classifiers | Although various techniques have been proposed to generate adversarial
samples for white-box attacks on text, little attention has been paid to
black-box attacks, which are more realistic scenarios. In this paper, we
present a novel algorithm, DeepWordBug, to effectively generate small text
perturbations in a black-box setting that forces a deep-learning classifier to
misclassify a text input. We employ novel scoring strategies to identify the
critical tokens that, if modified, cause the classifier to make an incorrect
prediction. Simple character-level transformations are applied to the
highest-ranked tokens in order to minimize the edit distance of the
perturbation, yet change the original classification. We evaluated DeepWordBug
on eight real-world text datasets, including text classification, sentiment
analysis, and spam detection. We compare the result of DeepWordBug with two
baselines: Random (Black-box) and Gradient (White-box). Our experimental
results indicate that DeepWordBug reduces the prediction accuracy of current
state-of-the-art deep-learning models, including a decrease of 68\% on average
for a Word-LSTM model and 48\% on average for a Char-CNN model.
|
Fairness in Supervised Learning: An Information Theoretic Approach | Automated decision making systems are increasingly being used in real-world
applications. In these systems for the most part, the decision rules are
derived by minimizing the training error on the available historical data.
Therefore, if there is a bias related to a sensitive attribute such as gender,
race, religion, etc. in the data, say, due to cultural/historical
discriminatory practices against a certain demographic, the system could
continue discrimination in decisions by including the said bias in its decision
rule. We present an information theoretic framework for designing fair
predictors from data, which aim to prevent discrimination against a specified
sensitive attribute in a supervised learning setting. We use equalized odds as
the criterion for discrimination, which demands that the prediction should be
independent of the protected attribute conditioned on the actual label. To
ensure fairness and generalization simultaneously, we compress the data to an
auxiliary variable, which is used for the prediction task. This auxiliary
variable is chosen such that it is decontaminated from the discriminatory
attribute in the sense of equalized odds. The final predictor is obtained by
applying a Bayesian decision rule to the auxiliary variable.
|
SuperNeurons: Dynamic GPU Memory Management for Training Deep Neural
Networks | Going deeper and wider in neural architectures improves the accuracy, while
the limited GPU DRAM places an undesired restriction on the network design
domain. Deep Learning (DL) practitioners either need change to less desired
network architectures, or nontrivially dissect a network across multiGPUs.
These distract DL practitioners from concentrating on their original machine
learning tasks. We present SuperNeurons: a dynamic GPU memory scheduling
runtime to enable the network training far beyond the GPU DRAM capacity.
SuperNeurons features 3 memory optimizations, \textit{Liveness Analysis},
\textit{Unified Tensor Pool}, and \textit{Cost-Aware Recomputation}, all
together they effectively reduce the network-wide peak memory usage down to the
maximal memory usage among layers. We also address the performance issues in
those memory saving techniques. Given the limited GPU DRAM, SuperNeurons not
only provisions the necessary memory for the training, but also dynamically
allocates the memory for convolution workspaces to achieve the high
performance. Evaluations against Caffe, Torch, MXNet and TensorFlow have
demonstrated that SuperNeurons trains at least 3.2432 deeper network than
current ones with the leading performance. Particularly, SuperNeurons can train
ResNet2500 that has $10^4$ basic network layers on a 12GB K40c.
|
Cost-Sensitive Convolution based Neural Networks for Imbalanced
Time-Series Classification | Some deep convolutional neural networks were proposed for time-series
classification and class imbalanced problems. However, those models performed
degraded and even failed to recognize the minority class of an imbalanced
temporal sequences dataset. Minority samples would bring troubles for temporal
deep learning classifiers due to the equal treatments of majority and minority
class. Until recently, there were few works applying deep learning on
imbalanced time-series classification (ITSC) tasks. Here, this paper aimed at
tackling ITSC problems with deep learning. An adaptive cost-sensitive learning
strategy was proposed to modify temporal deep learning models. Through the
proposed strategy, classifiers could automatically assign misclassification
penalties to each class. In the experimental section, the proposed method was
utilized to modify five neural networks. They were evaluated on a large volume,
real-life and imbalanced time-series dataset with six metrics. Each single
network was also tested alone and combined with several mainstream data
samplers. Experimental results illustrated that the proposed cost-sensitive
modified networks worked well on ITSC tasks. Compared to other methods, the
cost-sensitive convolution neural network and residual network won out in the
terms of all metrics. Consequently, the proposed cost-sensitive learning
strategy can be used to modify deep learning classifiers from cost-insensitive
to cost-sensitive. Those cost-sensitive convolutional networks can be
effectively applied to address ITSC issues.
|
A Survey on Compiler Autotuning using Machine Learning | Since the mid-1990s, researchers have been trying to use machine-learning
based approaches to solve a number of different compiler optimization problems.
These techniques primarily enhance the quality of the obtained results and,
more importantly, make it feasible to tackle two main compiler optimization
problems: optimization selection (choosing which optimizations to apply) and
phase-ordering (choosing the order of applying optimizations). The compiler
optimization space continues to grow due to the advancement of applications,
increasing number of compiler optimizations, and new target architectures.
Generic optimization passes in compilers cannot fully leverage newly introduced
optimizations and, therefore, cannot keep up with the pace of increasing
options. This survey summarizes and classifies the recent advances in using
machine learning for the compiler optimization field, particularly on the two
major problems of (1) selecting the best optimizations and (2) the
phase-ordering of optimizations. The survey highlights the approaches taken so
far, the obtained results, the fine-grain classification among different
approaches and finally, the influential papers of the field.
|
Which Training Methods for GANs do actually Converge? | Recent work has shown local convergence of GAN training for absolutely
continuous data and generator distributions. In this paper, we show that the
requirement of absolute continuity is necessary: we describe a simple yet
prototypical counterexample showing that in the more realistic case of
distributions that are not absolutely continuous, unregularized GAN training is
not always convergent. Furthermore, we discuss regularization strategies that
were recently proposed to stabilize GAN training. Our analysis shows that GAN
training with instance noise or zero-centered gradient penalties converges. On
the other hand, we show that Wasserstein-GANs and WGAN-GP with a finite number
of discriminator updates per generator update do not always converge to the
equilibrium point. We discuss these results, leading us to a new explanation
for the stability problems of GAN training. Based on our analysis, we extend
our convergence results to more general GANs and prove local convergence for
simplified gradient penalties even if the generator and data distribution lie
on lower dimensional manifolds. We find these penalties to work well in
practice and use them to learn high-resolution generative image models for a
variety of datasets with little hyperparameter tuning.
|
Towards a more efficient representation of imputation operators in TPOT | Automated Machine Learning encompasses a set of meta-algorithms intended to
design and apply machine learning techniques (e.g., model selection,
hyperparameter tuning, model assessment, etc.). TPOT, a software for optimizing
machine learning pipelines based on genetic programming (GP), is a novel
example of this kind of applications. Recently we have proposed a way to
introduce imputation methods as part of TPOT. While our approach was able to
deal with problems with missing data, it can produce a high number of
unfeasible pipelines. In this paper we propose a strongly-typed-GP based
approach that enforces constraint satisfaction by GP solutions. The enhancement
we introduce is based on the redefinition of the operators and implicit
enforcement of constraints in the generation of the GP trees. We evaluate the
method to introduce imputation methods as part of TPOT. We show that the method
can notably increase the efficiency of the GP search for optimal pipelines.
|
Can Computers Create Art? | This essay discusses whether computers, using Artificial Intelligence (AI),
could create art. First, the history of technologies that automated aspects of
art is surveyed, including photography and animation. In each case, there were
initial fears and denial of the technology, followed by a blossoming of new
creative and professional opportunities for artists. The current hype and
reality of Artificial Intelligence (AI) tools for art making is then discussed,
together with predictions about how AI tools will be used. It is then
speculated about whether it could ever happen that AI systems could be credited
with authorship of artwork. It is theorized that art is something created by
social agents, and so computers cannot be credited with authorship of art in
our current understanding. A few ways that this could change are also
hypothesized.
|
An Explicit Convergence Rate for Nesterov's Method from SDP | The framework of Integral Quadratic Constraints (IQC) introduced by Lessard
et al. (2014) reduces the computation of upper bounds on the convergence rate
of several optimization algorithms to semi-definite programming (SDP). In
particular, this technique was applied to Nesterov's accelerated method (NAM).
For quadratic functions, this SDP was explicitly solved leading to a new bound
on the convergence rate of NAM, and for arbitrary strongly convex functions it
was shown numerically that IQC can improve bounds from Nesterov (2004).
Unfortunately, an explicit analytic solution to the SDP was not provided. In
this paper, we provide such an analytical solution, obtaining a new general and
explicit upper bound on the convergence rate of NAM, which we further optimize
over its parameters. To the best of our knowledge, this is the best, and
explicit, upper bound on the convergence rate of NAM for strongly convex
functions.
|
Multivariate LSTM-FCNs for Time Series Classification | Over the past decade, multivariate time series classification has received
great attention. We propose transforming the existing univariate time series
classification models, the Long Short Term Memory Fully Convolutional Network
(LSTM-FCN) and Attention LSTM-FCN (ALSTM-FCN), into a multivariate time series
classification model by augmenting the fully convolutional block with a
squeeze-and-excitation block to further improve accuracy. Our proposed models
outperform most state-of-the-art models while requiring minimum preprocessing.
The proposed models work efficiently on various complex multivariate time
series classification tasks such as activity recognition or action recognition.
Furthermore, the proposed models are highly efficient at test time and small
enough to deploy on memory constrained systems.
|
Brain EEG Time Series Selection: A Novel Graph-Based Approach for
Classification | Brain Electroencephalography (EEG) classification is widely applied to
analyze cerebral diseases in recent years. Unfortunately, invalid/noisy EEGs
degrade the diagnosis performance and most previously developed methods ignore
the necessity of EEG selection for classification. To this end, this paper
proposes a novel maximum weight clique-based EEG selection approach, named
mwcEEGs, to map EEG selection to searching maximum similarity-weighted cliques
from an improved Fr\'{e}chet distance-weighted undirected EEG graph
simultaneously considering edge weights and vertex weights. Our mwcEEGs
improves the classification performance by selecting intra-clique pairwise
similar and inter-clique discriminative EEGs with similarity threshold
$\delta$. Experimental results demonstrate the algorithm effectiveness compared
with the state-of-the-art time series selection algorithms on real-world EEG
datasets.
|
Non-Parametric Transformation Networks | ConvNets, through their architecture, only enforce invariance to translation.
In this paper, we introduce a new class of deep convolutional architectures
called Non-Parametric Transformation Networks (NPTNs) which can learn
\textit{general} invariances and symmetries directly from data. NPTNs are a
natural generalization of ConvNets and can be optimized directly using gradient
descent. Unlike almost all previous works in deep architectures, they make no
assumption regarding the structure of the invariances present in the data and
in that aspect are flexible and powerful. We also model ConvNets and NPTNs
under a unified framework called Transformation Networks (TN), which yields a
better understanding of the connection between the two. We demonstrate the
efficacy of NPTNs on data such as MNIST with extreme transformations and
CIFAR10 where they outperform baselines, and further outperform several recent
algorithms on ETH-80. They do so while having the same number of parameters. We
also show that they are more effective than ConvNets in modelling symmetries
and invariances from data, without the explicit knowledge of the added
arbitrary nuisance transformations. Finally, we replace ConvNets with NPTNs
within Capsule Networks and show that this enables Capsule Nets to perform even
better.
|
Fix your classifier: the marginal value of training the last weight
layer | Neural networks are commonly used as models for classification for a wide
variety of tasks. Typically, a learned affine transformation is placed at the
end of such models, yielding a per-class value used for classification. This
classifier can have a vast number of parameters, which grows linearly with the
number of possible classes, thus requiring increasingly more resources. In this
work we argue that this classifier can be fixed, up to a global scale constant,
with little or no loss of accuracy for most tasks, allowing memory and
computational benefits. Moreover, we show that by initializing the classifier
with a Hadamard matrix we can speed up inference as well. We discuss the
implications for current understanding of neural network models.
|
Evaluation of Machine Learning Fameworks on Finis Terrae II | Machine Learning (ML) and Deep Learning (DL) are two technologies used to
extract representations of the data for a specific purpose. ML algorithms take
a set of data as input to generate one or several predictions. To define the
final version of one model, usually there is an initial step devoted to train
the algorithm (get the right final values of the parameters of the model).
There are several techniques, from supervised learning to reinforcement
learning, which have different requirements. On the market, there are some
frameworks or APIs that reduce the effort for designing a new ML model. In this
report, using the benchmark DLBENCH, we will analyse the performance and the
execution modes of some well-known ML frameworks on the Finis Terrae II
supercomputer when supervised learning is used. The report will show that
placement of data and allocated hardware can have a large influence on the
final timeto-solution.
|
DCDistance: A Supervised Text Document Feature extraction based on class
labels | Text Mining is a field that aims at extracting information from textual data.
One of the challenges of such field of study comes from the pre-processing
stage in which a vector (and structured) representation should be extracted
from unstructured data. The common extraction creates large and sparse vectors
representing the importance of each term to a document. As such, this usually
leads to the curse-of-dimensionality that plagues most machine learning
algorithms. To cope with this issue, in this paper we propose a new supervised
feature extraction and reduction algorithm, named DCDistance, that creates
features based on the distance between a document to a representative of each
class label. As such, the proposed technique can reduce the features set in
more than 99% of the original set. Additionally, this algorithm was also
capable of improving the classification accuracy over a set of benchmark
datasets when compared to traditional and state-of-the-art features selection
algorithms.
|
Deep Reinforcement Learning of Cell Movement in the Early Stage of C.
elegans Embryogenesis | Cell movement in the early phase of C. elegans development is regulated by a
highly complex process in which a set of rules and connections are formulated
at distinct scales. Previous efforts have shown that agent-based, multi-scale
modeling systems can integrate physical and biological rules and provide new
avenues to study developmental systems. However, the application of these
systems to model cell movement is still challenging and requires a
comprehensive understanding of regulation networks at the right scales. Recent
developments in deep learning and reinforcement learning provide an
unprecedented opportunity to explore cell movement using 3D time-lapse images.
We present a deep reinforcement learning approach within an ABM system to
characterize cell movement in C. elegans embryogenesis. Our modeling system
captures the complexity of cell movement patterns in the embryo and overcomes
the local optimization problem encountered by traditional rule-based, ABM that
uses greedy algorithms. We tested our model with two real developmental
processes: the anterior movement of the Cpaaa cell via intercalation and the
rearrangement of the left-right asymmetry. In the first case, model results
showed that Cpaaa's intercalation is an active directional cell movement caused
by the continuous effects from a longer distance, as opposed to a passive
movement caused by neighbor cell movements. This is because the learning-based
simulation found that a passive movement model could not lead Cpaaa to the
predefined destination. In the second case, a leader-follower mechanism well
explained the collective cell movement pattern. These results showed that our
approach to introduce deep reinforcement learning into ABM can test regulatory
mechanisms by exploring cell migration paths in a reverse engineering
perspective. This model opens new doors to explore large datasets generated by
live imaging.
|
Towards Imperceptible and Robust Adversarial Example Attacks against
Neural Networks | Machine learning systems based on deep neural networks, being able to produce
state-of-the-art results on various perception tasks, have gained mainstream
adoption in many applications. However, they are shown to be vulnerable to
adversarial example attack, which generates malicious output by adding slight
perturbations to the input. Previous adversarial example crafting methods,
however, use simple metrics to evaluate the distances between the original
examples and the adversarial ones, which could be easily detected by human
eyes. In addition, these attacks are often not robust due to the inevitable
noises and deviation in the physical world. In this work, we present a new
adversarial example attack crafting method, which takes the human perceptual
system into consideration and maximizes the noise tolerance of the crafted
adversarial example. Experimental results demonstrate the efficacy of the
proposed technique.
|
Sparsity-based Defense against Adversarial Attacks on Linear Classifiers | Deep neural networks represent the state of the art in machine learning in a
growing number of fields, including vision, speech and natural language
processing. However, recent work raises important questions about the
robustness of such architectures, by showing that it is possible to induce
classification errors through tiny, almost imperceptible, perturbations.
Vulnerability to such "adversarial attacks", or "adversarial examples", has
been conjectured to be due to the excessive linearity of deep networks. In this
paper, we study this phenomenon in the setting of a linear classifier, and show
that it is possible to exploit sparsity in natural data to combat
$\ell_{\infty}$-bounded adversarial perturbations. Specifically, we demonstrate
the efficacy of a sparsifying front end via an ensemble averaged analysis, and
experimental results for the MNIST handwritten digit database. To the best of
our knowledge, this is the first work to show that sparsity provides a
theoretically rigorous framework for defense against adversarial attacks.
|
tau-FPL: Tolerance-Constrained Learning in Linear Time | Learning a classifier with control on the false-positive rate plays a
critical role in many machine learning applications. Existing approaches either
introduce prior knowledge dependent label cost or tune parameters based on
traditional classifiers, which lack consistency in methodology because they do
not strictly adhere to the false-positive rate constraint. In this paper, we
propose a novel scoring-thresholding approach, tau-False Positive Learning
(tau-FPL) to address this problem. We show the scoring problem which takes the
false-positive rate tolerance into accounts can be efficiently solved in linear
time, also an out-of-bootstrap thresholding method can transform the learned
ranking function into a low false-positive classifier. Both theoretical
analysis and experimental results show superior performance of the proposed
tau-FPL over existing approaches.
|
Predicting Movie Genres Based on Plot Summaries | This project explores several Machine Learning methods to predict movie
genres based on plot summaries. Naive Bayes, Word2Vec+XGBoost and Recurrent
Neural Networks are used for text classification, while K-binary
transformation, rank method and probabilistic classification with learned
probability threshold are employed for the multi-label problem involved in the
genre tagging task.Experiments with more than 250,000 movies show that
employing the Gated Recurrent Units (GRU) neural networks for the probabilistic
classification with learned probability threshold approach achieves the best
result on the test set. The model attains a Jaccard Index of 50.0%, a F-score
of 0.56, and a hit rate of 80.5%.
|
Improving Orbit Prediction Accuracy through Supervised Machine Learning | Due to the lack of information such as the space environment condition and
resident space objects' (RSOs') body characteristics, current orbit predictions
that are solely grounded on physics-based models may fail to achieve required
accuracy for collision avoidance and have led to satellite collisions already.
This paper presents a methodology to predict RSOs' trajectories with higher
accuracy than that of the current methods. Inspired by the machine learning
(ML) theory through which the models are learned based on large amounts of
observed data and the prediction is conducted without explicitly modeling space
objects and space environment, the proposed ML approach integrates
physics-based orbit prediction algorithms with a learning-based process that
focuses on reducing the prediction errors. Using a simulation-based space
catalog environment as the test bed, the paper demonstrates three types of
generalization capability for the proposed ML approach: 1) the ML model can be
used to improve the same RSO's orbit information that is not available during
the learning process but shares the same time interval as the training data; 2)
the ML model can be used to improve predictions of the same RSO at future
epochs; and 3) the ML model based on a RSO can be applied to other RSOs that
share some common features.
|
Unsupervised Cipher Cracking Using Discrete GANs | This work details CipherGAN, an architecture inspired by CycleGAN used for
inferring the underlying cipher mapping given banks of unpaired ciphertext and
plaintext. We demonstrate that CipherGAN is capable of cracking language data
enciphered using shift and Vigenere ciphers to a high degree of fidelity and
for vocabularies much larger than previously achieved. We present how CycleGAN
can be made compatible with discrete data and train in a stable way. We then
prove that the technique used in CipherGAN avoids the common problem of
uninformative discrimination associated with GANs applied to discrete data.
|
Leapfrogging for parallelism in deep neural networks | We present a technique, which we term leapfrogging, to parallelize back-
propagation in deep neural networks. We show that this technique yields a
savings of $1-1/k$ of a dominant term in backpropagation, where k is the number
of threads (or gpus).
|
Generalizing, Decoding, and Optimizing Support Vector Machine
Classification | The classification of complex data usually requires the composition of
processing steps. Here, a major challenge is the selection of optimal
algorithms for preprocessing and classification (including parameterizations).
Nowadays, parts of the optimization process are automized but expert knowledge
and manual work are still required. We present three steps to face this process
and ease the optimization. Namely, we take a theoretical view on classical
classifiers, provide an approach to interpret the classifier together with the
preprocessing, and integrate both into one framework which enables a
semiautomatic optimization of the processing chain and which interfaces
numerous algorithms.
|
Topic Modeling on Health Journals with Regularized Variational Inference | Topic modeling enables exploration and compact representation of a corpus.
The CaringBridge (CB) dataset is a massive collection of journals written by
patients and caregivers during a health crisis. Topic modeling on the CB
dataset, however, is challenging due to the asynchronous nature of multiple
authors writing about their health journeys. To overcome this challenge we
introduce the Dynamic Author-Persona topic model (DAP), a probabilistic
graphical model designed for temporal corpora with multiple authors. The
novelty of the DAP model lies in its representation of authors by a persona ---
where personas capture the propensity to write about certain topics over time.
Further, we present a regularized variational inference algorithm, which we use
to encourage the DAP model's personas to be distinct. Our results show
significant improvements over competing topic models --- particularly after
regularization, and highlight the DAP model's unique ability to capture common
journeys shared by different authors.
|
On the Complexity of the Weighted Fused Lasso | The solution path of the 1D fused lasso for an $n$-dimensional input is
piecewise linear with $\mathcal{O}(n)$ segments (Hoefling et al. 2010 and
Tibshirani et al 2011). However, existing proofs of this bound do not hold for
the weighted fused lasso. At the same time, results for the generalized lasso,
of which the weighted fused lasso is a special case, allow $\Omega(3^n)$
segments (Mairal et al. 2012). In this paper, we prove that the number of
segments in the solution path of the weighted fused lasso is
$\mathcal{O}(n^2)$, and that, for some instances, it is $\Omega(n^2)$. We also
give a new, very simple, proof of the $\mathcal{O}(n)$ bound for the fused
lasso.
|
An octree cells occupancy geometric dimensionality descriptor for
massive on-server point cloud visualisation and classification | Lidar datasets are becoming more and more common. They are appreciated for
their precise 3D nature, and have a wide range of applications, such as surface
reconstruction, object detection, visualisation, etc. For all this
applications, having additional semantic information per point has potential of
increasing the quality and the efficiency of the application. In the last
decade the use of Machine Learning and more specifically classification methods
have proved to be successful to create this semantic information. In this
paradigm, the goal is to classify points into a set of given classes (for
instance tree, building, ground, other). Some of these methods use descriptors
(also called feature) of a point to learn and predict its class. Designing the
descriptors is then the heart of these methods. Descriptors can be based on
points geometry and attributes, use contextual information, etc. Furthermore,
descriptors can be used by humans for easier visual understanding and sometimes
filtering. In this work we propose a new simple geometric descriptor that gives
information about the implicit local dimensionality of the point cloud at
various scale. For instance a tree seen from afar is more volumetric in nature
(3D), yet locally each leaves is rather planar (2D). To do so we build an
octree centred on the point to consider, and compare the variation of the
occupancy of the cells across the levels of the octree. We compare this
descriptor with the state of the art dimensionality descriptor and show its
interest. We further test the descriptor for classification within the Point
Cloud Server, and demonstrate efficiency and correctness results.
|
Global Convergence of Policy Gradient Methods for the Linear Quadratic
Regulator | Direct policy gradient methods for reinforcement learning and continuous
control problems are a popular approach for a variety of reasons: 1) they are
easy to implement without explicit knowledge of the underlying model 2) they
are an "end-to-end" approach, directly optimizing the performance metric of
interest 3) they inherently allow for richly parameterized policies. A notable
drawback is that even in the most basic continuous control problem (that of
linear quadratic regulators), these methods must solve a non-convex
optimization problem, where little is understood about their efficiency from
both computational and statistical perspectives. In contrast, system
identification and model based planning in optimal control theory have a much
more solid theoretical footing, where much is known with regards to their
computational and statistical properties. This work bridges this gap showing
that (model free) policy gradient methods globally converge to the optimal
solution and are efficient (polynomially so in relevant problem dependent
quantities) with regards to their sample and computational complexities.
|
Multi-Label Learning from Medical Plain Text with Convolutional Residual
Models | Predicting diagnoses from Electronic Health Records (EHRs) is an important
medical application of multi-label learning. We propose a convolutional
residual model for multi-label classification from doctor notes in EHR data. A
given patient may have multiple diagnoses, and therefore multi-label learning
is required. We employ a Convolutional Neural Network (CNN) to encode plain
text into a fixed-length sentence embedding vector. Since diagnoses are
typically correlated, a deep residual network is employed on top of the CNN
encoder, to capture label (diagnosis) dependencies and incorporate information
directly from the encoded sentence vector. A real EHR dataset is considered,
and we compare the proposed model with several well-known baselines, to predict
diagnoses based on doctor notes. Experimental results demonstrate the
superiority of the proposed convolutional residual model.
|
Understanding the Disharmony between Dropout and Batch Normalization by
Variance Shift | This paper first answers the question "why do the two most powerful
techniques Dropout and Batch Normalization (BN) often lead to a worse
performance when they are combined together?" in both theoretical and
statistical aspects. Theoretically, we find that Dropout would shift the
variance of a specific neural unit when we transfer the state of that network
from train to test. However, BN would maintain its statistical variance, which
is accumulated from the entire learning procedure, in the test phase. The
inconsistency of that variance (we name this scheme as "variance shift") causes
the unstable numerical behavior in inference that leads to more erroneous
predictions finally, when applying Dropout before BN. Thorough experiments on
DenseNet, ResNet, ResNeXt and Wide ResNet confirm our findings. According to
the uncovered mechanism, we next explore several strategies that modifies
Dropout and try to overcome the limitations of their combination by avoiding
the variance shift risks.
|
GitGraph - Architecture Search Space Creation through Frequent
Computational Subgraph Mining | The dramatic success of deep neural networks across multiple application
areas often relies on experts painstakingly designing a network architecture
specific to each task. To simplify this process and make it more accessible, an
emerging research effort seeks to automate the design of neural network
architectures, using e.g. evolutionary algorithms or reinforcement learning or
simple search in a constrained space of neural modules.
Considering the typical size of the search space (e.g. $10^{10}$ candidates
for a $10$-layer network) and the cost of evaluating a single candidate,
current architecture search methods are very restricted. They either rely on
static pre-built modules to be recombined for the task at hand, or they define
a static hand-crafted framework within which they can generate new
architectures from the simplest possible operations.
In this paper, we relax these restrictions, by capitalizing on the collective
wisdom contained in the plethora of neural networks published in online code
repositories. Concretely, we (a) extract and publish GitGraph, a corpus of
neural architectures and their descriptions; (b) we create problem-specific
neural architecture search spaces, implemented as a textual search mechanism
over GitGraph; (c) we propose a method of identifying unique common subgraphs
within the architectures solving each problem (e.g., image processing,
reinforcement learning), that can then serve as modules in the newly created
problem specific neural search space.
|
MORF: A Framework for Predictive Modeling and Replication At Scale With
Privacy-Restricted MOOC Data | Big data repositories from online learning platforms such as Massive Open
Online Courses (MOOCs) represent an unprecedented opportunity to advance
research on education at scale and impact a global population of learners. To
date, such research has been hindered by poor reproducibility and a lack of
replication, largely due to three types of barriers: experimental, inferential,
and data. We present a novel system for large-scale computational research, the
MOOC Replication Framework (MORF), to jointly address these barriers. We
discuss MORF's architecture, an open-source platform-as-a-service (PaaS) which
includes a simple, flexible software API providing for multiple modes of
research (predictive modeling or production rule analysis) integrated with a
high-performance computing environment. All experiments conducted on MORF use
executable Docker containers which ensure complete reproducibility while
allowing for the use of any software or language which can be installed in the
linux-based Docker container. Each experimental artifact is assigned a DOI and
made publicly available. MORF has the potential to accelerate and democratize
research on its massive data repository, which currently includes over 200
MOOCs, as demonstrated by initial research conducted on the platform. We also
highlight ways in which MORF represents a solution template to a more general
class of problems faced by computational researchers in other domains.
|
Rank Selection of CP-decomposed Convolutional Layers with Variational
Bayesian Matrix Factorization | Convolutional Neural Networks (CNNs) is one of successful method in many
areas such as image classification tasks. However, the amount of memory and
computational cost needed for CNNs inference obstructs them to run efficiently
in mobile devices because of memory and computational ability limitation. One
of the method to compress CNNs is compressing the layers iteratively, i.e. by
layer-by-layer compression and fine-tuning, with CP-decomposition in
convolutional layers. To compress with CP-decomposition, rank selection is
important. In the previous approach rank selection that is based on sensitivity
of each layer, the average rank of the network was still arbitrarily selected.
Additionally, the rank of all layers were decided before whole process of
iterative compression, while the rank of a layer can be changed after
fine-tuning. Therefore, this paper proposes selecting rank of each layer using
Variational Bayesian Matrix Factorization (VBMF) which is more systematic than
arbitrary approach. Furthermore, to consider the change of each layer's rank
after fine-tuning of previous iteration, the method is applied just before
compressing the target layer, i.e. after fine-tuning of the previous iteration.
The results show better accuracy while also having more compression rate in
AlexNet's convolutional layers compression.
|
Neural Feature Learning From Relational Database | Feature engineering is one of the most important but most tedious tasks in
data science. This work studies automation of feature learning from relational
database. We first prove theoretically that finding the optimal features from
relational data for predictive tasks is NP-hard. We propose an efficient
rule-based approach based on heuristics and a deep neural network to
automatically learn appropriate features from relational data. We benchmark our
approaches in ensembles in past Kaggle competitions. Our new approach wins late
medals and beats the state-of-the-art solutions with significant margins. To
the best of our knowledge, this is the first time an automated data science
system could win medals in Kaggle competitions with complex relational
database.
|
StressedNets: Efficient Feature Representations via Stress-induced
Evolutionary Synthesis of Deep Neural Networks | The computational complexity of leveraging deep neural networks for
extracting deep feature representations is a significant barrier to its
widespread adoption, particularly for use in embedded devices. One particularly
promising strategy to addressing the complexity issue is the notion of
evolutionary synthesis of deep neural networks, which was demonstrated to
successfully produce highly efficient deep neural networks while retaining
modeling performance. Here, we further extend upon the evolutionary synthesis
strategy for achieving efficient feature extraction via the introduction of a
stress-induced evolutionary synthesis framework, where stress signals are
imposed upon the synapses of a deep neural network during training to induce
stress and steer the synthesis process towards the production of more efficient
deep neural networks over successive generations and improved model fidelity at
a greater efficiency. The proposed stress-induced evolutionary synthesis
approach is evaluated on a variety of different deep neural network
architectures (LeNet5, AlexNet, and YOLOv2) on different tasks (object
classification and object detection) to synthesize efficient StressedNets over
multiple generations. Experimental results demonstrate the efficacy of the
proposed framework to synthesize StressedNets with significant improvement in
network architecture efficiency (e.g., 40x for AlexNet and 33x for YOLOv2) and
speed improvements (e.g., 5.5x inference speed-up for YOLOv2 on an Nvidia Tegra
X1 mobile processor).
|
Time Series Segmentation through Automatic Feature Learning | Internet of things (IoT) applications have become increasingly popular in
recent years, with applications ranging from building energy monitoring to
personal health tracking and activity recognition. In order to leverage these
data, automatic knowledge extraction - whereby we map from observations to
interpretable states and transitions - must be done at scale. As such, we have
seen many recent IoT data sets include annotations with a human expert
specifying states, recorded as a set of boundaries and associated labels in a
data sequence. These data can be used to build automatic labeling algorithms
that produce labels as an expert would. Here, we refer to human-specified
boundaries as breakpoints. Traditional changepoint detection methods only look
for statistically-detectable boundaries that are defined as abrupt variations
in the generative parameters of a data sequence. However, we observe that
breakpoints occur on more subtle boundaries that are non-trivial to detect with
these statistical methods. In this work, we propose a new unsupervised
approach, based on deep learning, that outperforms existing techniques and
learns the more subtle, breakpoint boundaries with a high accuracy. Through
extensive experiments on various real-world data sets - including
human-activity sensing data, speech signals, and electroencephalogram (EEG)
activity traces - we demonstrate the effectiveness of our algorithm for
practical applications. Furthermore, we show that our approach achieves
significantly better performance than previous methods.
|
On the Direction of Discrimination: An Information-Theoretic Analysis of
Disparate Impact in Machine Learning | In the context of machine learning, disparate impact refers to a form of
systematic discrimination whereby the output distribution of a model depends on
the value of a sensitive attribute (e.g., race or gender). In this paper, we
propose an information-theoretic framework to analyze the disparate impact of a
binary classification model. We view the model as a fixed channel, and quantify
disparate impact as the divergence in output distributions over two groups. Our
aim is to find a correction function that can perturb the input distributions
of each group to align their output distributions. We present an optimization
problem that can be solved to obtain a correction function that will make the
output distributions statistically indistinguishable. We derive closed-form
expressions to efficiently compute the correction function, and demonstrate the
benefits of our framework on a recidivism prediction problem based on the
ProPublica COMPAS dataset.
|
Deep Canonically Correlated LSTMs | We examine Deep Canonically Correlated LSTMs as a way to learn nonlinear
transformations of variable length sequences and embed them into a correlated,
fixed dimensional space. We use LSTMs to transform multi-view time-series data
non-linearly while learning temporal relationships within the data. We then
perform correlation analysis on the outputs of these neural networks to find a
correlated subspace through which we get our final representation via
projection. This work follows from previous work done on Deep Canonical
Correlation (DCCA), in which deep feed-forward neural networks were used to
learn nonlinear transformations of data while maximizing correlation.
|
Expectation Propagation for Approximate Inference: Free Probability
Framework | We study asymptotic properties of expectation propagation (EP) -- a method
for approximate inference originally developed in the field of machine
learning. Applied to generalized linear models, EP iteratively computes a
multivariate Gaussian approximation to the exact posterior distribution. The
computational complexity of the repeated update of covariance matrices severely
limits the application of EP to large problem sizes. In this study, we present
a rigorous analysis by means of free probability theory that allows us to
overcome this computational bottleneck if specific data matrices in the problem
fulfill certain properties of asymptotic freeness. We demonstrate the relevance
of our approach on the gene selection problem of a microarray dataset.
|
Combinatorial Preconditioners for Proximal Algorithms on Graphs | We present a novel preconditioning technique for proximal optimization
methods that relies on graph algorithms to construct effective preconditioners.
Such combinatorial preconditioners arise from partitioning the graph into
forests. We prove that certain decompositions lead to a theoretically optimal
condition number. We also show how ideal decompositions can be realized using
matroid partitioning and propose efficient greedy variants thereof for
large-scale problems. Coupled with specialized solvers for the resulting scaled
proximal subproblems, the preconditioned algorithm achieves competitive
performance in machine learning and vision applications.
|
A Comparative Study of Rule Extraction for Recurrent Neural Networks | Understanding recurrent networks through rule extraction has a long history.
This has taken on new interests due to the need for interpreting or verifying
neural networks. One basic form for representing stateful rules is
deterministic finite automata (DFA). Previous research shows that extracting
DFAs from trained second-order recurrent networks is not only possible but also
relatively stable. Recently, several new types of recurrent networks with more
complicated architectures have been introduced. These handle challenging
learning tasks usually involving sequential data. However, it remains an open
problem whether DFAs can be adequately extracted from these models.
Specifically, it is not clear how DFA extraction will be affected when applied
to different recurrent networks trained on data sets with different levels of
complexity. Here, we investigate DFA extraction on several widely adopted
recurrent networks that are trained to learn a set of seven regular Tomita
grammars. We first formally analyze the complexity of Tomita grammars and
categorize these grammars according to that complexity. Then we empirically
evaluate different recurrent networks for their performance of DFA extraction
on all Tomita grammars. Our experiments show that for most recurrent networks,
their extraction performance decreases as the complexity of the underlying
grammar increases. On grammars of lower complexity, most recurrent networks
obtain desirable extraction performance. As for grammars with the highest level
of complexity, while several complicated models fail with only certain
recurrent networks having satisfactory extraction performance.
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.