title
stringlengths 7
246
| abstract
stringlengths 6
3.31k
|
|---|---|
Efficient Rank Aggregation via Lehmer Codes | We propose a novel rank aggregation method based on converting permutations
into their corresponding Lehmer codes or other subdiagonal images. Lehmer
codes, also known as inversion vectors, are vector representations of
permutations in which each coordinate can take values not restricted by the
values of other coordinates. This transformation allows for decoupling of the
coordinates and for performing aggregation via simple scalar median or mode
computations. We present simulation results illustrating the performance of
this completely parallelizable approach and analytically prove that both the
mode and median aggregation procedure recover the correct centroid aggregate
with small sample complexity when the permutations are drawn according to the
well-known Mallows models. The proposed Lehmer code approach may also be used
on partial rankings, with similar performance guarantees.
|
Skip Connections Eliminate Singularities | Skip connections made the training of very deep networks possible and have
become an indispensable component in a variety of neural architectures. A
completely satisfactory explanation for their success remains elusive. Here, we
present a novel explanation for the benefits of skip connections in training
very deep networks. The difficulty of training deep networks is partly due to
the singularities caused by the non-identifiability of the model. Several such
singularities have been identified in previous works: (i) overlap singularities
caused by the permutation symmetry of nodes in a given layer, (ii) elimination
singularities corresponding to the elimination, i.e. consistent deactivation,
of nodes, (iii) singularities generated by the linear dependence of the nodes.
These singularities cause degenerate manifolds in the loss landscape that slow
down learning. We argue that skip connections eliminate these singularities by
breaking the permutation symmetry of nodes, by reducing the possibility of node
elimination and by making the nodes less linearly dependent. Moreover, for
typical initializations, skip connections move the network away from the
"ghosts" of these singularities and sculpt the landscape around them to
alleviate the learning slow-down. These hypotheses are supported by evidence
from simplified models, as well as from experiments with deep networks trained
on real-world datasets.
|
A Dirichlet Mixture Model of Hawkes Processes for Event Sequence
Clustering | We propose an effective method to solve the event sequence clustering
problems based on a novel Dirichlet mixture model of a special but significant
type of point processes --- Hawkes process. In this model, each event sequence
belonging to a cluster is generated via the same Hawkes process with specific
parameters, and different clusters correspond to different Hawkes processes.
The prior distribution of the Hawkes processes is controlled via a Dirichlet
distribution. We learn the model via a maximum likelihood estimator (MLE) and
propose an effective variational Bayesian inference algorithm. We specifically
analyze the resulting EM-type algorithm in the context of inner-outer
iterations and discuss several inner iteration allocation strategies. The
identifiability of our model, the convergence of our learning method, and its
sample complexity are analyzed in both theoretical and empirical ways, which
demonstrate the superiority of our method to other competitors. The proposed
method learns the number of clusters automatically and is robust to model
misspecification. Experiments on both synthetic and real-world data show that
our method can learn diverse triggering patterns hidden in asynchronous event
sequences and achieve encouraging performance on clustering purity and
consistency.
|
Learning the distribution with largest mean: two bandit frameworks | Over the past few years, the multi-armed bandit model has become increasingly
popular in the machine learning community, partly because of applications
including online content optimization. This paper reviews two different
sequential learning tasks that have been considered in the bandit literature ;
they can be formulated as (sequentially) learning which distribution has the
highest mean among a set of distributions, with some constraints on the
learning process. For both of them (regret minimization and best arm
identification) we present recent, asymptotically optimal algorithms. We
compare the behaviors of the sampling rule of each algorithm as well as the
complexity terms associated to each problem.
|
Towards "AlphaChem": Chemical Synthesis Planning with Tree Search and
Deep Neural Network Policies | Retrosynthesis is a technique to plan the chemical synthesis of organic
molecules, for example drugs, agro- and fine chemicals. In retrosynthesis, a
search tree is built by analysing molecules recursively and dissecting them
into simpler molecular building blocks until one obtains a set of known
building blocks. The search space is intractably large, and it is difficult to
determine the value of retrosynthetic positions. Here, we propose to model
retrosynthesis as a Markov Decision Process. In combination with a Deep Neural
Network policy learned from essentially the complete published knowledge of
chemistry, Monte Carlo Tree Search (MCTS) can be used to evaluate positions. In
exploratory studies, we demonstrate that MCTS with neural network policies
outperforms the traditionally used best-first search with hand-coded
heuristics.
|
Representation of big data by dimension reduction | Suppose the data consist of a set $S$ of points $x_j, 1 \leq j \leq J$,
distributed in a bounded domain $D \subset R^N$, where $N$ and $J$ are large
numbers. In this paper an algorithm is proposed for checking whether there
exists a manifold $\mathbb{M}$ of low dimension near which many of the points
of $S$ lie and finding such $\mathbb{M}$ if it exists. There are many dimension
reduction algorithms, both linear and non-linear. Our algorithm is simple to
implement and has some advantages compared with the known algorithms. If there
is a manifold of low dimension near which most of the data points lie, the
proposed algorithm will find it. Some numerical results are presented
illustrating the algorithm and analyzing its performance compared to the
classical PCA (principal component analysis) and Isomap.
|
On orthogonality and learning recurrent networks with long term
dependencies | It is well known that it is challenging to train deep neural networks and
recurrent neural networks for tasks that exhibit long term dependencies. The
vanishing or exploding gradient problem is a well known issue associated with
these challenges. One approach to addressing vanishing and exploding gradients
is to use either soft or hard constraints on weight matrices so as to encourage
or enforce orthogonality. Orthogonal matrices preserve gradient norm during
backpropagation and may therefore be a desirable property. This paper explores
issues with optimization convergence, speed and gradient stability when
encouraging or enforcing orthogonality. To perform this analysis, we propose a
weight matrix factorization and parameterization strategy through which we can
bound matrix norms and therein control the degree of expansivity induced during
backpropagation. We find that hard constraints on orthogonality can negatively
affect the speed of convergence and model performance.
|
Stochastic Graphlet Embedding | Graph-based methods are known to be successful in many machine learning and
pattern classification tasks. These methods consider semi-structured data as
graphs where nodes correspond to primitives (parts, interest points, segments,
etc.) and edges characterize the relationships between these primitives.
However, these non-vectorial graph data cannot be straightforwardly plugged
into off-the-shelf machine learning algorithms without a preliminary step of --
explicit/implicit -- graph vectorization and embedding. This embedding process
should be resilient to intra-class graph variations while being highly
discriminant. In this paper, we propose a novel high-order stochastic graphlet
embedding (SGE) that maps graphs into vector spaces. Our main contribution
includes a new stochastic search procedure that efficiently parses a given
graph and extracts/samples unlimitedly high-order graphlets. We consider these
graphlets, with increasing orders, to model local primitives as well as their
increasingly complex interactions. In order to build our graph representation,
we measure the distribution of these graphlets into a given graph, using
particular hash functions that efficiently assign sampled graphlets into
isomorphic sets with a very low probability of collision. When combined with
maximum margin classifiers, these graphlet-based representations have positive
impact on the performance of pattern comparison and recognition as corroborated
through extensive experiments using standard benchmark databases.
|
PCA-Initialized Deep Neural Networks Applied To Document Image Analysis | In this paper, we present a novel approach for initializing deep neural
networks, i.e., by turning PCA into neural layers. Usually, the initialization
of the weights of a deep neural network is done in one of the three following
ways: 1) with random values, 2) layer-wise, usually as Deep Belief Network or
as auto-encoder, and 3) re-use of layers from another network (transfer
learning). Therefore, typically, many training epochs are needed before
meaningful weights are learned, or a rather similar dataset is required for
seeding a fine-tuning of transfer learning. In this paper, we describe how to
turn a PCA into an auto-encoder, by generating an encoder layer of the PCA
parameters and furthermore adding a decoding layer. We analyze the
initialization technique on real documents. First, we show that a PCA-based
initialization is quick and leads to a very stable initialization. Furthermore,
for the task of layout analysis we investigate the effectiveness of PCA-based
initialization and show that it outperforms state-of-the-art random weight
initialization methods.
|
On the Futility of Learning Complex Frame-Level Language Models for
Chord Recognition | Chord recognition systems use temporal models to post-process frame-wise
chord preditions from acoustic models. Traditionally, first-order models such
as Hidden Markov Models were used for this task, with recent works suggesting
to apply Recurrent Neural Networks instead. Due to their ability to learn
longer-term dependencies, these models are supposed to learn and to apply
musical knowledge, instead of just smoothing the output of the acoustic model.
In this paper, we argue that learning complex temporal models at the level of
audio frames is futile on principle, and that non-Markovian models do not
perform better than their first-order counterparts. We support our argument
through three experiments on the McGill Billboard dataset. The first two show
1) that when learning complex temporal models at the frame level, improvements
in chord sequence modelling are marginal; and 2) that these improvements do not
translate when applied within a full chord recognition system. The third, still
rather preliminary experiment gives first indications that the use of complex
sequential models for chord prediction at higher temporal levels might be more
promising.
|
Communication-Optimal Distributed Clustering | Clustering large datasets is a fundamental problem with a number of
applications in machine learning. Data is often collected on different sites
and clustering needs to be performed in a distributed manner with low
communication. We would like the quality of the clustering in the distributed
setting to match that in the centralized setting for which all the data resides
on a single site. In this work, we study both graph and geometric clustering
problems in two distributed models: (1) a point-to-point model, and (2) a model
with a broadcast channel. We give protocols in both models which we show are
nearly optimal by proving almost matching communication lower bounds. Our work
highlights the surprising power of a broadcast channel for clustering problems;
roughly speaking, to spectrally cluster $n$ points or $n$ vertices in a graph
distributed across $s$ servers, for a worst-case partitioning the communication
complexity in a point-to-point model is $n \cdot s$, while in the broadcast
model it is $n + s$. A similar phenomenon holds for the geometric setting as
well. We implement our algorithms and demonstrate this phenomenon on real life
datasets, showing that our algorithms are also very efficient in practice.
|
Machine learning based compact photonic structure design for strong
light confinement | We present a novel approach based on machine learning for designing photonic
structures. In particular, we focus on strong light confinement that allows the
design of an efficient free-space-to-waveguide coupler which is made of Si-
slab overlying on the top of silica substrate. The learning algorithm is
implemented using bitwise square Si- cells and the whole optimized device has a
footprint of $\boldsymbol{2 \, \mu m \times 1\, \mu m}$, which is the smallest
size ever achieved numerically. To find the effect of Si- slab thickness on the
sub-wavelength focusing and strong coupling characteristics of optimized
photonic structure, we carried out three-dimensional time-domain numerical
calculations. Corresponding optimum values of full width at half maximum and
coupling efficiency were calculated as $\boldsymbol{0.158 \lambda}$ and
$\boldsymbol{-1.87\,dB}$ with slab thickness of $\boldsymbol{280nm}$. Compared
to the conventional counterparts, the optimized lens and coupler designs are
easy-to-fabricate via optical lithography techniques, quite compact, and can
operate at telecommunication wavelengths. The outcomes of the presented study
show that machine learning can be beneficial for efficient photonic designs in
various potential applications such as polarization-division, beam manipulation
and optical interconnects.
|
On SGD's Failure in Practice: Characterizing and Overcoming Stalling | Stochastic Gradient Descent (SGD) is widely used in machine learning problems
to efficiently perform empirical risk minimization, yet, in practice, SGD is
known to stall before reaching the actual minimizer of the empirical risk. SGD
stalling has often been attributed to its sensitivity to the conditioning of
the problem; however, as we demonstrate, SGD will stall even when applied to a
simple linear regression problem with unity condition number for standard
learning rates. Thus, in this work, we numerically demonstrate and
mathematically argue that stalling is a crippling and generic limitation of SGD
and its variants in practice. Once we have established the problem of stalling,
we generalize an existing framework for hedging against its effects, which (1)
deters SGD and its variants from stalling, (2) still provides convergence
guarantees, and (3) makes SGD and its variants more practical methods for
minimization.
|
Generative Adversarial Networks recover features in astrophysical images
of galaxies beyond the deconvolution limit | Observations of astrophysical objects such as galaxies are limited by various
sources of random and systematic noise from the sky background, the optical
system of the telescope and the detector used to record the data. Conventional
deconvolution techniques are limited in their ability to recover features in
imaging data by the Shannon-Nyquist sampling theorem. Here we train a
generative adversarial network (GAN) on a sample of $4,550$ images of nearby
galaxies at $0.01<z<0.02$ from the Sloan Digital Sky Survey and conduct
$10\times$ cross validation to evaluate the results. We present a method using
a GAN trained on galaxy images that can recover features from artificially
degraded images with worse seeing and higher noise than the original with a
performance which far exceeds simple deconvolution. The ability to better
recover detailed features such as galaxy morphology from low-signal-to-noise
and low angular resolution imaging data significantly increases our ability to
study existing data sets of astrophysical objects as well as future
observations with observatories such as the Large Synoptic Sky Telescope (LSST)
and the Hubble and James Webb space telescopes.
|
Convergence Results for Neural Networks via Electrodynamics | We study whether a depth two neural network can learn another depth two
network using gradient descent. Assuming a linear output node, we show that the
question of whether gradient descent converges to the target function is
equivalent to the following question in electrodynamics: Given $k$ fixed
protons in $\mathbb{R}^d,$ and $k$ electrons, each moving due to the attractive
force from the protons and repulsive force from the remaining electrons,
whether at equilibrium all the electrons will be matched up with the protons,
up to a permutation. Under the standard electrical force, this follows from the
classic Earnshaw's theorem. In our setting, the force is determined by the
activation function and the input distribution. Building on this equivalence,
we prove the existence of an activation function such that gradient descent
learns at least one of the hidden nodes in the target network. Iterating, we
show that gradient descent can be used to learn the entire network one node at
a time.
|
Algorithmic Performance-Accuracy Trade-off in 3D Vision Applications
Using HyperMapper | In this paper we investigate an emerging application, 3D scene understanding,
likely to be significant in the mobile space in the near future. The goal of
this exploration is to reduce execution time while meeting our quality of
result objectives. In previous work we showed for the first time that it is
possible to map this application to power constrained embedded systems,
highlighting that decision choices made at the algorithmic design-level have
the most impact.
As the algorithmic design space is too large to be exhaustively evaluated, we
use a previously introduced multi-objective Random Forest Active Learning
prediction framework dubbed HyperMapper, to find good algorithmic designs. We
show that HyperMapper generalizes on a recent cutting edge 3D scene
understanding algorithm and on a modern GPU-based computer architecture.
HyperMapper is able to beat an expert human hand-tuning the algorithmic
parameters of the class of Computer Vision applications taken under
consideration in this paper automatically. In addition, we use crowd-sourcing
using a 3D scene understanding Android app to show that the Pareto front
obtained on an embedded system can be used to accelerate the same application
on all the 83 smart-phones and tablets crowd-sourced with speedups ranging from
2 to over 12.
|
Segmentation of optic disc, fovea and retinal vasculature using a single
convolutional neural network | We have developed and trained a convolutional neural network to automatically
and simultaneously segment optic disc, fovea and blood vessels. Fundus images
were normalised before segmentation was performed to enforce consistency in
background lighting and contrast. For every effective point in the fundus
image, our algorithm extracted three channels of input from the neighbourhood
of the point and forward the response across the 7 layer network. In average,
our segmentation achieved an accuracy of 92.68 percent on the testing set from
Drive database.
|
Recovering True Classifier Performance in Positive-Unlabeled Learning | A common approach in positive-unlabeled learning is to train a classification
model between labeled and unlabeled data. This strategy is in fact known to
give an optimal classifier under mild conditions; however, it results in biased
empirical estimates of the classifier performance. In this work, we show that
the typically used performance measures such as the receiver operating
characteristic curve, or the precision-recall curve obtained on such data can
be corrected with the knowledge of class priors; i.e., the proportions of the
positive and negative examples in the unlabeled data. We extend the results to
a noisy setting where some of the examples labeled positive are in fact
negative and show that the correction also requires the knowledge of the
proportion of noisy examples in the labeled positives. Using state-of-the-art
algorithms to estimate the positive class prior and the proportion of noise, we
experimentally evaluate two correction approaches and demonstrate their
efficacy on real-life data.
|
Deep Learning the Indus Script | Standardized corpora of undeciphered scripts, a necessary starting point for
computational epigraphy, requires laborious human effort for their preparation
from raw archaeological records. Automating this process through machine
learning algorithms can be of significant aid to epigraphical research. Here,
we take the first steps in this direction and present a deep learning pipeline
that takes as input images of the undeciphered Indus script, as found in
archaeological artifacts, and returns as output a string of graphemes, suitable
for inclusion in a standard corpus. The image is first decomposed into regions
using Selective Search and these regions are classified as containing textual
and/or graphical information using a convolutional neural network. Regions
classified as potentially containing text are hierarchically merged and trimmed
to remove non-textual information. The remaining textual part of the image is
segmented using standard image processing techniques to isolate individual
graphemes. This set is finally passed to a second convolutional neural network
to classify the graphemes, based on a standard corpus. The classifier can
identify the presence or absence of the most frequent Indus grapheme, the "jar"
sign, with an accuracy of 92%. Our results demonstrate the great potential of
deep learning approaches in computational epigraphy and, more generally, in the
digital humanities.
|
Optimal Schemes for Discrete Distribution Estimation under Locally
Differential Privacy | We consider the minimax estimation problem of a discrete distribution with
support size $k$ under privacy constraints. A privatization scheme is applied
to each raw sample independently, and we need to estimate the distribution of
the raw samples from the privatized samples. A positive number $\epsilon$
measures the privacy level of a privatization scheme. For a given $\epsilon,$
we consider the problem of constructing optimal privatization schemes with
$\epsilon$-privacy level, i.e., schemes that minimize the expected estimation
loss for the worst-case distribution. Two schemes in the literature provide
order optimal performance in the high privacy regime where $\epsilon$ is very
close to $0,$ and in the low privacy regime where $e^{\epsilon}\approx k,$
respectively.
In this paper, we propose a new family of schemes which substantially improve
the performance of the existing schemes in the medium privacy regime when $1\ll
e^{\epsilon} \ll k.$ More concretely, we prove that when $3.8 < \epsilon
<\ln(k/9) ,$ our schemes reduce the expected estimation loss by $50\%$ under
$\ell_2^2$ metric and by $30\%$ under $\ell_1$ metric over the existing
schemes. We also prove a lower bound for the region $e^{\epsilon} \ll k,$ which
implies that our schemes are order optimal in this regime.
|
IQN: An Incremental Quasi-Newton Method with Local Superlinear
Convergence Rate | The problem of minimizing an objective that can be written as the sum of a
set of $n$ smooth and strongly convex functions is considered. The Incremental
Quasi-Newton (IQN) method proposed here belongs to the family of stochastic and
incremental methods that have a cost per iteration independent of $n$. IQN
iterations are a stochastic version of BFGS iterations that use memory to
reduce the variance of stochastic approximations. The convergence properties of
IQN bridge a gap between deterministic and stochastic quasi-Newton methods.
Deterministic quasi-Newton methods exploit the possibility of approximating the
Newton step using objective gradient differences. They are appealing because
they have a smaller computational cost per iteration relative to Newton's
method and achieve a superlinear convergence rate under customary regularity
assumptions. Stochastic quasi-Newton methods utilize stochastic gradient
differences in lieu of actual gradient differences. This makes their
computational cost per iteration independent of the number of objective
functions $n$. However, existing stochastic quasi-Newton methods have sublinear
or linear convergence at best. IQN is the first stochastic quasi-Newton method
proven to converge superlinearly in a local neighborhood of the optimal
solution. IQN differs from state-of-the-art incremental quasi-Newton methods in
three aspects: (i) The use of aggregated information of variables, gradients,
and quasi-Newton Hessian approximation matrices to reduce the noise of gradient
and Hessian approximations. (ii) The approximation of each individual function
by its Taylor's expansion in which the linear and quadratic terms are evaluated
with respect to the same iterate. (iii) The use of a cyclic scheme to update
the functions in lieu of a random selection routine. We use these fundamental
properties of IQN to establish its local superlinear convergence rate.
|
HashNet: Deep Learning to Hash by Continuation | Learning to hash has been widely applied to approximate nearest neighbor
search for large-scale multimedia retrieval, due to its computation efficiency
and retrieval quality. Deep learning to hash, which improves retrieval quality
by end-to-end representation learning and hash encoding, has received
increasing attention recently. Subject to the ill-posed gradient difficulty in
the optimization with sign activations, existing deep learning to hash methods
need to first learn continuous representations and then generate binary hash
codes in a separated binarization step, which suffer from substantial loss of
retrieval quality. This work presents HashNet, a novel deep architecture for
deep learning to hash by continuation method with convergence guarantees, which
learns exactly binary hash codes from imbalanced similarity data. The key idea
is to attack the ill-posed gradient problem in optimizing deep networks with
non-smooth binary activations by continuation method, in which we begin from
learning an easier network with smoothed activation function and let it evolve
during the training, until it eventually goes back to being the original,
difficult to optimize, deep network with the sign activation function.
Comprehensive empirical evidence shows that HashNet can generate exactly binary
hash codes and yield state-of-the-art multimedia retrieval performance on
standard benchmarks.
|
Natasha: Faster Non-Convex Stochastic Optimization Via Strongly
Non-Convex Parameter | Given a nonconvex function that is an average of $n$ smooth functions, we
design stochastic first-order methods to find its approximate stationary
points. The convergence of our new methods depends on the smallest (negative)
eigenvalue $-\sigma$ of the Hessian, a parameter that describes how nonconvex
the function is.
Our methods outperform known results for a range of parameter $\sigma$, and
can be used to find approximate local minima. Our result implies an interesting
dichotomy: there exists a threshold $\sigma_0$ so that the currently fastest
methods for $\sigma>\sigma_0$ and for $\sigma<\sigma_0$ have different
behaviors: the former scales with $n^{2/3}$ and the latter scales with
$n^{3/4}$.
|
Pixel Recursive Super Resolution | We present a pixel recursive super resolution model that synthesizes
realistic details into images while enhancing their resolution. A low
resolution image may correspond to multiple plausible high resolution images,
thus modeling the super resolution process with a pixel independent conditional
model often results in averaging different details--hence blurry edges. By
contrast, our model is able to represent a multimodal conditional distribution
by properly modeling the statistical dependencies among the high resolution
image pixels, conditioned on a low resolution input. We employ a PixelCNN
architecture to define a strong prior over natural images and jointly optimize
this prior with a deep conditioning convolutional network. Human evaluations
indicate that samples from our proposed model look more photo realistic than a
strong L2 regression baseline.
|
An Introduction to Deep Learning for the Physical Layer | We present and discuss several novel applications of deep learning for the
physical layer. By interpreting a communications system as an autoencoder, we
develop a fundamental new way to think about communications system design as an
end-to-end reconstruction task that seeks to jointly optimize transmitter and
receiver components in a single process. We show how this idea can be extended
to networks of multiple transmitters and receivers and present the concept of
radio transformer networks as a means to incorporate expert domain knowledge in
the machine learning model. Lastly, we demonstrate the application of
convolutional neural networks on raw IQ samples for modulation classification
which achieves competitive accuracy with respect to traditional schemes relying
on expert features. The paper is concluded with a discussion of open challenges
and areas for future investigation.
|
Recurrent Neural Networks for anomaly detection in the Post-Mortem time
series of LHC superconducting magnets | This paper presents a model based on Deep Learning algorithms of LSTM and GRU
for facilitating an anomaly detection in Large Hadron Collider superconducting
magnets. We used high resolution data available in Post Mortem database to
train a set of models and chose the best possible set of their
hyper-parameters. Using Deep Learning approach allowed to examine a vast body
of data and extract the fragments which require further experts examination and
are regarded as anomalies. The presented method does not require tedious manual
threshold setting and operator attention at the stage of the system setup.
Instead, the automatic approach is proposed, which achieves according to our
experiments accuracy of 99%. This is reached for the largest dataset of 302 MB
and the following architecture of the network: single layer LSTM, 128 cells, 20
epochs of training, look_back=16, look_ahead=128, grid=100 and optimizer Adam.
All the experiments were run on GPU Nvidia Tesla K80
|
Structured Attention Networks | Attention networks have proven to be an effective approach for embedding
categorical inference within a deep neural network. However, for many tasks we
may want to model richer structural dependencies without abandoning end-to-end
training. In this work, we experiment with incorporating richer structural
distributions, encoded using graphical models, within deep networks. We show
that these structured attention networks are simple extensions of the basic
attention procedure, and that they allow for extending attention beyond the
standard soft-selection approach, such as attending to partial segmentations or
to subtrees. We experiment with two different classes of structured attention
networks: a linear-chain conditional random field and a graph-based parsing
model, and describe how these models can be practically implemented as neural
network layers. Experiments show that this approach is effective for
incorporating structural biases, and structured attention networks outperform
baseline attention models on a variety of synthetic and real tasks: tree
transduction, neural machine translation, question answering, and natural
language inference. We further find that models trained in this way learn
interesting unsupervised hidden representations that generalize simple
attention.
|
Deep Learning with Low Precision by Half-wave Gaussian Quantization | The problem of quantizing the activations of a deep neural network is
considered. An examination of the popular binary quantization approach shows
that this consists of approximating a classical non-linearity, the hyperbolic
tangent, by two functions: a piecewise constant sign function, which is used in
feedforward network computations, and a piecewise linear hard tanh function,
used in the backpropagation step during network learning. The problem of
approximating the ReLU non-linearity, widely used in the recent deep learning
literature, is then considered. An half-wave Gaussian quantizer (HWGQ) is
proposed for forward approximation and shown to have efficient implementation,
by exploiting the statistics of of network activations and batch normalization
operations commonly used in the literature. To overcome the problem of gradient
mismatch, due to the use of different forward and backward approximations,
several piece-wise backward approximators are then investigated. The
implementation of the resulting quantized network, denoted as HWGQ-Net, is
shown to achieve much closer performance to full precision networks, such as
AlexNet, ResNet, GoogLeNet and VGG-Net, than previously available low-precision
networks, with 1-bit binary weights and 2-bit quantized activations.
|
Intrinsic Grassmann Averages for Online Linear, Robust and Nonlinear
Subspace Learning | Principal Component Analysis (PCA) and Kernel Principal Component Analysis
(KPCA) are fundamental methods in machine learning for dimensionality
reduction. The former is a technique for finding this approximation in finite
dimensions and the latter is often in an infinite dimensional Reproducing
Kernel Hilbert-space (RKHS). In this paper, we present a geometric framework
for computing the principal linear subspaces in both situations as well as for
the robust PCA case, that amounts to computing the intrinsic average on the
space of all subspaces: the Grassmann manifold. Points on this manifold are
defined as the subspaces spanned by $K$-tuples of observations. The intrinsic
Grassmann average of these subspaces are shown to coincide with the principal
components of the observations when they are drawn from a Gaussian
distribution. We show similar results in the RKHS case and provide an efficient
algorithm for computing the projection onto the this average subspace. The
result is a method akin to KPCA which is substantially faster. Further, we
present a novel online version of the KPCA using our geometric framework.
Competitive performance of all our algorithms are demonstrated on a variety of
real and synthetic data sets.
|
Uncertainty-Aware Reinforcement Learning for Collision Avoidance | Reinforcement learning can enable complex, adaptive behavior to be learned
automatically for autonomous robotic platforms. However, practical deployment
of reinforcement learning methods must contend with the fact that the training
process itself can be unsafe for the robot. In this paper, we consider the
specific case of a mobile robot learning to navigate an a priori unknown
environment while avoiding collisions. In order to learn collision avoidance,
the robot must experience collisions at training time. However, high-speed
collisions, even at training time, could damage the robot. A successful
learning method must therefore proceed cautiously, experiencing only low-speed
collisions until it gains confidence. To this end, we present an
uncertainty-aware model-based learning algorithm that estimates the probability
of collision together with a statistical estimate of uncertainty. By
formulating an uncertainty-dependent cost function, we show that the algorithm
naturally chooses to proceed cautiously in unfamiliar environments, and
increases the velocity of the robot in settings where it has high confidence.
Our predictive model is based on bootstrapped neural networks using dropout,
allowing it to process raw sensory inputs from high-bandwidth sensors such as
cameras. Our experimental evaluation demonstrates that our method effectively
minimizes dangerous collisions at training time in an obstacle avoidance task
for a simulated and real-world quadrotor, and a real-world RC car. Videos of
the experiments can be found at https://sites.google.com/site/probcoll.
|
Fuzzy Clustering Data Given on the Ordinal Scale Based on Membership and
Likelihood Functions Sharing | A task of clustering data given in the ordinal scale under conditions of
overlapping clusters has been considered. It's proposed to use an approach
based on memberhsip and likelihood functions sharing. A number of performed
experiments proved effectiveness of the proposed method. The proposed method is
characterized by robustness to outliers due to a way of ordering values while
constructing membership functions.
|
Traffic Lights with Auction-Based Controllers: Algorithms and Real-World
Data | Real-time optimization of traffic flow addresses important practical
problems: reducing a driver's wasted time, improving city-wide efficiency,
reducing gas emissions and improving air quality. Much of the current research
in traffic-light optimization relies on extending the capabilities of traffic
lights to either communicate with each other or communicate with vehicles.
However, before such capabilities become ubiquitous, opportunities exist to
improve traffic lights by being more responsive to current traffic situations
within the current, already deployed, infrastructure. In this paper, we
introduce a traffic light controller that employs bidding within micro-auctions
to efficiently incorporate traffic sensor information; no other outside sources
of information are assumed. We train and test traffic light controllers on
large-scale data collected from opted-in Android cell-phone users over a period
of several months in Mountain View, California and the River North neighborhood
of Chicago, Illinois. The learned auction-based controllers surpass (in both
the relevant metrics of road-capacity and mean travel time) the currently
deployed lights, optimized static-program lights, and longer-term planning
approaches, in both cities, measured using real user driving data.
|
A Theoretical Analysis of First Heuristics of Crowdsourced Entity
Resolution | Entity resolution (ER) is the task of identifying all records in a database
that refer to the same underlying entity, and are therefore duplicates of each
other. Due to inherent ambiguity of data representation and poor data quality,
ER is a challenging task for any automated process. As a remedy, human-powered
ER via crowdsourcing has become popular in recent years. Using crowd to answer
queries is costly and time consuming. Furthermore, crowd-answers can often be
faulty. Therefore, crowd-based ER methods aim to minimize human participation
without sacrificing the quality and use a computer generated similarity matrix
actively. While, some of these methods perform well in practice, no theoretical
analysis exists for them, and further their worst case performances do not
reflect the experimental findings. This creates a disparity in the
understanding of the popular heuristics for this problem. In this paper, we
make the first attempt to close this gap. We provide a thorough analysis of the
prominent heuristic algorithms for crowd-based ER. We justify experimental
observations with our analysis and information theoretic lower bounds.
|
Towards Better Analysis of Machine Learning Models: A Visual Analytics
Perspective | Interactive model analysis, the process of understanding, diagnosing, and
refining a machine learning model with the help of interactive visualization,
is very important for users to efficiently solve real-world artificial
intelligence and data mining problems. Dramatic advances in big data analytics
has led to a wide variety of interactive model analysis tasks. In this paper,
we present a comprehensive analysis and interpretation of this rapidly
developing area. Specifically, we classify the relevant work into three
categories: understanding, diagnosis, and refinement. Each category is
exemplified by recent influential work. Possible future research opportunities
are also explored and discussed.
|
A Learning-Based Approach for Lane Departure Warning Systems with a
Personalized Driver Model | Misunderstanding of driver correction behaviors (DCB) is the primary reason
for false warnings of lane-departure-prediction systems. We propose a
learning-based approach to predicting unintended lane-departure behaviors (LDB)
and the chance for drivers to bring the vehicle back to the lane. First, in
this approach, a personalized driver model for lane-departure and lane-keeping
behavior is established by combining the Gaussian mixture model and the hidden
Markov model. Second, based on this model, we develop an online model-based
prediction algorithm to predict the forthcoming vehicle trajectory and judge
whether the driver will demonstrate an LDB or a DCB. We also develop a warning
strategy based on the model-based prediction algorithm that allows the
lane-departure warning system to be acceptable for drivers according to the
predicted trajectory. In addition, the naturalistic driving data of 10 drivers
is collected through the University of Michigan Safety Pilot Model Deployment
program to train the personalized driver model and validate this approach. We
compare the proposed method with a basic time-to-lane-crossing (TLC) method and
a TLC-directional sequence of piecewise lateral slopes (TLC-DSPLS) method. The
results show that the proposed approach can reduce the false-warning rate to
3.07\%.
|
Simple to Complex Cross-modal Learning to Rank | The heterogeneity-gap between different modalities brings a significant
challenge to multimedia information retrieval. Some studies formalize the
cross-modal retrieval tasks as a ranking problem and learn a shared multi-modal
embedding space to measure the cross-modality similarity. However, previous
methods often establish the shared embedding space based on linear mapping
functions which might not be sophisticated enough to reveal more complicated
inter-modal correspondences. Additionally, current studies assume that the
rankings are of equal importance, and thus all rankings are used
simultaneously, or a small number of rankings are selected randomly to train
the embedding space at each iteration. Such strategies, however, always suffer
from outliers as well as reduced generalization capability due to their lack of
insightful understanding of procedure of human cognition. In this paper, we
involve the self-paced learning theory with diversity into the cross-modal
learning to rank and learn an optimal multi-modal embedding space based on
non-linear mapping functions. This strategy enhances the model's robustness to
outliers and achieves better generalization via training the model gradually
from easy rankings by diverse queries to more complex ones. An efficient
alternative algorithm is exploited to solve the proposed challenging problem
with fast convergence in practice. Extensive experimental results on several
benchmark datasets indicate that the proposed method achieves significant
improvements over the state-of-the-arts in this literature.
|
Network-based methods for outcome prediction in the "sample space" | In this thesis we present the novel semi-supervised network-based algorithm
P-Net, which is able to rank and classify patients with respect to a specific
phenotype or clinical outcome under study. The peculiar and innovative
characteristic of this method is that it builds a network of samples/patients,
where the nodes represent the samples and the edges are functional or genetic
relationships between individuals (e.g. similarity of expression profiles), to
predict the phenotype under study. In other words, it constructs the network in
the "sample space" and not in the "biomarker space" (where nodes represent
biomolecules (e.g. genes, proteins) and edges represent functional or genetic
relationships between nodes), as usual in state-of-the-art methods. To assess
the performances of P-Net, we apply it on three different publicly available
datasets from patients afflicted with a specific type of tumor: pancreatic
cancer, melanoma and ovarian cancer dataset, by using the data and following
the experimental set-up proposed in two recently published papers [Barter et
al., 2014, Winter et al., 2012]. We show that network-based methods in the
"sample space" can achieve results competitive with classical supervised
inductive systems. Moreover, the graph representation of the samples can be
easily visualized through networks and can be used to gain visual clues about
the relationships between samples, taking into account the phenotype associated
or predicted for each sample. To our knowledge this is one of the first works
that proposes graph-based algorithms working in the "sample space" of the
biomolecular profiles of the patients to predict their phenotype or outcome,
thus contributing to a novel research line in the framework of the Network
Medicine.
|
Latent Hinge-Minimax Risk Minimization for Inference from a Small Number
of Training Samples | Deep Learning (DL) methods show very good performance when trained on large,
balanced data sets. However, many practical problems involve imbalanced data
sets, or/and classes with a small number of training samples. The performance
of DL methods as well as more traditional classifiers drops significantly in
such settings. Most of the existing solutions for imbalanced problems focus on
customizing the data for training. A more principled solution is to use mixed
Hinge-Minimax risk [19] specifically designed to solve binary problems with
imbalanced training sets. Here we propose a Latent Hinge Minimax (LHM) risk and
a training algorithm that generalizes this paradigm to an ensemble of
hyperplanes that can form arbitrary complex, piecewise linear boundaries. To
extract good features, we combine LHM model with CNN via transfer learning. To
solve multi-class problem we map pre-trained category-specific LHM classifiers
to a multi-class neural network and adjust the weights with very fast tuning.
LHM classifier enables the use of unlabeled data in its training and the
mapping allows for multi-class inference, resulting in a classifier that
performs better than alternatives when trained on a small number of training
samples.
|
Cluster-based Kriging Approximation Algorithms for Complexity Reduction | Kriging or Gaussian Process Regression is applied in many fields as a
non-linear regression model as well as a surrogate model in the field of
evolutionary computation. However, the computational and space complexity of
Kriging, that is cubic and quadratic in the number of data points respectively,
becomes a major bottleneck with more and more data available nowadays. In this
paper, we propose a general methodology for the complexity reduction, called
cluster Kriging, where the whole data set is partitioned into smaller clusters
and multiple Kriging models are built on top of them. In addition, four Kriging
approximation algorithms are proposed as candidate algorithms within the new
framework. Each of these algorithms can be applied to much larger data sets
while maintaining the advantages and power of Kriging. The proposed algorithms
are explained in detail and compared empirically against a broad set of
existing state-of-the-art Kriging approximation methods on a well-defined
testing framework. According to the empirical study, the proposed algorithms
consistently outperform the existing algorithms. Moreover, some practical
suggestions are provided for using the proposed algorithms.
|
An Experimental Study of Deep Convolutional Features For Iris
Recognition | Iris is one of the popular biometrics that is widely used for identity
authentication. Different features have been used to perform iris recognition
in the past. Most of them are based on hand-crafted features designed by
biometrics experts. Due to tremendous success of deep learning in computer
vision problems, there has been a lot of interest in applying features learned
by convolutional neural networks on general image recognition to other tasks
such as segmentation, face recognition, and object detection. In this paper, we
have investigated the application of deep features extracted from VGG-Net for
iris recognition. The proposed scheme has been tested on two well-known iris
databases, and has shown promising results with the best accuracy rate of
99.4\%, which outperforms the previous best result.
|
Deep learning and the Schr\"odinger equation | We have trained a deep (convolutional) neural network to predict the
ground-state energy of an electron in four classes of confining two-dimensional
electrostatic potentials. On randomly generated potentials, for which there is
no analytic form for either the potential or the ground-state energy, the
neural network model was able to predict the ground-state energy to within
chemical accuracy, with a median absolute error of 1.49 mHa. We also
investigate the performance of the model in predicting other quantities such as
the kinetic energy and the first excited-state energy of random potentials.
|
A scikit-based Python environment for performing multi-label
classification | scikit-multilearn is a Python library for performing multi-label
classification. The library is compatible with the scikit/scipy ecosystem and
uses sparse matrices for all internal operations. It provides native Python
implementations of popular multi-label classification methods alongside a novel
framework for label space partitioning and division. It includes modern
algorithm adaptation methods, network-based label space division approaches,
which extracts label dependency information and multi-label embedding
classifiers. It provides python wrapped access to the extensive multi-label
method stack from Java libraries and makes it possible to extend deep learning
single-label methods for multi-label tasks. The library allows multi-label
stratification and data set management. The implementation is more efficient in
problem transformation than other established libraries, has good test coverage
and follows PEP8. Source code and documentation can be downloaded from
http://scikit.ml and also via pip. The library follows BSD licensing scheme.
|
Optimizing Cost-Sensitive SVM for Imbalanced Data :Connecting Cluster to
Classification | Class imbalance is one of the challenging problems for machine learning in
many real-world applications, such as coal and gas burst accident monitoring:
the burst premonition data is extreme smaller than the normal data, however,
which is the highlight we truly focus on. Cost-sensitive adjustment approach is
a typical algorithm-level method resisting the data set imbalance. For SVMs
classifier, which is modified to incorporate varying penalty parameter(C) for
each of considered groups of examples. However, the C value is determined
empirically, or is calculated according to the evaluation metric, which need to
be computed iteratively and time consuming. This paper presents a novel
cost-sensitive SVM method whose penalty parameter C optimized on the basis of
cluster probability density function(PDF) and the cluster PDF is estimated only
according to similarity matrix and some predefined hyper-parameters.
Experimental results on various standard benchmark data sets and real-world
data with different ratios of imbalance show that the proposed method is
effective in comparison with commonly used cost-sensitive techniques.
|
Calibrating Energy-based Generative Adversarial Networks | In this paper, we propose to equip Generative Adversarial Networks with the
ability to produce direct energy estimates for samples.Specifically, we propose
a flexible adversarial training framework, and prove this framework not only
ensures the generator converges to the true data distribution, but also enables
the discriminator to retain the density information at the global optimal. We
derive the analytic form of the induced solution, and analyze the properties.
In order to make the proposed framework trainable in practice, we introduce two
effective approximation techniques. Empirically, the experiment results closely
match our theoretical analysis, verifying the discriminator is able to recover
the energy of data distribution.
|
Search Intelligence: Deep Learning For Dominant Category Prediction | Deep Neural Networks, and specifically fully-connected convolutional neural
networks are achieving remarkable results across a wide variety of domains.
They have been trained to achieve state-of-the-art performance when applied to
problems such as speech recognition, image classification, natural language
processing and bioinformatics. Most of these deep learning models when applied
to classification employ the softmax activation function for prediction and aim
to minimize cross-entropy loss. In this paper, we have proposed a supervised
model for dominant category prediction to improve search recall across all eBay
classifieds platforms. The dominant category label for each query in the last
90 days is first calculated by summing the total number of collaborative clicks
among all categories. The category having the highest number of collaborative
clicks for the given query will be considered its dominant category. Second,
each query is transformed to a numeric vector by mapping each unique word in
the query document to a unique integer value; all padded to equal length based
on the maximum document length within the pre-defined vocabulary size. A
fully-connected deep convolutional neural network (CNN) is then applied for
classification. The proposed model achieves very high classification accuracy
compared to other state-of-the-art machine learning techniques.
|
Toward the automated analysis of complex diseases in genome-wide
association studies using genetic programming | Machine learning has been gaining traction in recent years to meet the demand
for tools that can efficiently analyze and make sense of the ever-growing
databases of biomedical data in health care systems around the world. However,
effectively using machine learning methods requires considerable domain
expertise, which can be a barrier of entry for bioinformaticians new to
computational data science methods. Therefore, off-the-shelf tools that make
machine learning more accessible can prove invaluable for bioinformaticians. To
this end, we have developed an open source pipeline optimization tool
(TPOT-MDR) that uses genetic programming to automatically design machine
learning pipelines for bioinformatics studies. In TPOT-MDR, we implement
Multifactor Dimensionality Reduction (MDR) as a feature construction method for
modeling higher-order feature interactions, and combine it with a new expert
knowledge-guided feature selector for large biomedical data sets. We
demonstrate TPOT-MDR's capabilities using a combination of simulated and real
world data sets from human genetics and find that TPOT-MDR significantly
outperforms modern machine learning methods such as logistic regression and
eXtreme Gradient Boosting (XGBoost). We further analyze the best pipeline
discovered by TPOT-MDR for a real world problem and highlight TPOT-MDR's
ability to produce a high-accuracy solution that is also easily interpretable.
|
Predicting Pairwise Relations with Neural Similarity Encoders | Matrix factorization is at the heart of many machine learning algorithms, for
example, dimensionality reduction (e.g. kernel PCA) or recommender systems
relying on collaborative filtering. Understanding a singular value
decomposition (SVD) of a matrix as a neural network optimization problem
enables us to decompose large matrices efficiently while dealing naturally with
missing values in the given matrix. But most importantly, it allows us to learn
the connection between data points' feature vectors and the matrix containing
information about their pairwise relations. In this paper we introduce a novel
neural network architecture termed Similarity Encoder (SimEc), which is
designed to simultaneously factorize a given target matrix while also learning
the mapping to project the data points' feature vectors into a similarity
preserving embedding space. This makes it possible to, for example, easily
compute out-of-sample solutions for new data points. Additionally, we
demonstrate that SimEc can preserve non-metric similarities and even predict
multiple pairwise relations between data points at once.
|
Neural Discourse Structure for Text Categorization | We show that discourse structure, as defined by Rhetorical Structure Theory
and provided by an existing discourse parser, benefits text categorization. Our
approach uses a recursive neural network and a newly proposed attention
mechanism to compute a representation of the text that focuses on salient
content, from the perspective of both RST and the task. Experiments consider
variants of the approach and illustrate its strengths and weaknesses.
|
Low Rank Matrix Recovery with Simultaneous Presence of Outliers and
Sparse Corruption | We study a data model in which the data matrix D can be expressed as D = L +
S + C, where L is a low rank matrix, S an element-wise sparse matrix and C a
matrix whose non-zero columns are outlying data points. To date, robust PCA
algorithms have solely considered models with either S or C, but not both. As
such, existing algorithms cannot account for simultaneous element-wise and
column-wise corruptions. In this paper, a new robust PCA algorithm that is
robust to simultaneous types of corruption is proposed. Our approach hinges on
the sparse approximation of a sparsely corrupted column so that the sparse
expansion of a column with respect to the other data points is used to
distinguish a sparsely corrupted inlier column from an outlying data point. We
also develop a randomized design which provides a scalable implementation of
the proposed approach. The core idea of sparse approximation is analyzed
analytically where we show that the underlying ell_1-norm minimization can
obtain the representation of an inlier in presence of sparse corruptions.
|
A multi-channel approach for automatic microseismic event association
using RANSAC-based arrival time event clustering(RATEC) | In the presence of background noise, arrival times picked from a surface
microseismic data set usually include a number of false picks that can lead to
uncertainty in location estimation. To eliminate false picks and improve the
accuracy of location estimates, we develop an association algorithm termed
RANSAC-based Arrival Time Event Clustering (RATEC) that clusters picked arrival
times into event groups based on random sampling and fitting moveout curves
that approximate hyperbolas. Arrival times far from the fitted hyperbolas are
classified as false picks and removed from the data set prior to location
estimation. Simulations of synthetic data for a 1-D linear array show that
RATEC is robust under different noise conditions and generally applicable to
various types of subsurface structures. By generalizing the underlying moveout
model, RATEC is extended to the case of a 2-D surface monitoring array. The
effectiveness of event location for the 2-D case is demonstrated using a data
set collected by the 5200-element dense Long Beach array. The obtained results
suggest that RATEC is effective in removing false picks and hence can be used
for phase association before location estimates.
|
Sparse Algorithm for Robust LSSVM in Primal Space | As enjoying the closed form solution, least squares support vector machine
(LSSVM) has been widely used for classification and regression problems having
the comparable performance with other types of SVMs. However, LSSVM has two
drawbacks: sensitive to outliers and lacking sparseness. Robust LSSVM (R-LSSVM)
overcomes the first partly via nonconvex truncated loss function, but the
current algorithms for R-LSSVM with the dense solution are faced with the
second drawback and are inefficient for training large-scale problems. In this
paper, we interpret the robustness of R-LSSVM from a re-weighted viewpoint and
give a primal R-LSSVM by the representer theorem. The new model may have sparse
solution if the corresponding kernel matrix has low rank. Then approximating
the kernel matrix by a low-rank matrix and smoothing the loss function by
entropy penalty function, we propose a convergent sparse R-LSSVM (SR-LSSVM)
algorithm to achieve the sparse solution of primal R-LSSVM, which overcomes two
drawbacks of LSSVM simultaneously. The proposed algorithm has lower complexity
than the existing algorithms and is very efficient for training large-scale
problems. Many experimental results illustrate that SR-LSSVM can achieve better
or comparable performance with less training time than related algorithms,
especially for training large scale problems.
|
Continuous-Time User Modeling in the Presence of Badges: A Probabilistic
Approach | User modeling plays an important role in delivering customized web services
to the users and improving their engagement. However, most user models in the
literature do not explicitly consider the temporal behavior of users. More
recently, continuous-time user modeling has gained considerable attention and
many user behavior models have been proposed based on temporal point processes.
However, typical point process based models often considered the impact of peer
influence and content on the user participation and neglected other factors.
Gamification elements, are among those factors that are neglected, while they
have a strong impact on user participation in online services. In this paper,
we propose interdependent multi-dimensional temporal point processes that
capture the impact of badges on user participation besides the peer influence
and content factors. We extend the proposed processes to model user actions
over the community based question and answering websites, and propose an
inference algorithm based on Variational-EM that can efficiently learn the
model parameters. Extensive experiments on both synthetic and real data
gathered from Stack Overflow show that our inference algorithm learns the
parameters efficiently and the proposed method can better predict the user
behavior compared to the alternatives.
|
Representations of language in a model of visually grounded speech
signal | We present a visually grounded model of speech perception which projects
spoken utterances and images to a joint semantic space. We use a multi-layer
recurrent highway network to model the temporal nature of spoken speech, and
show that it learns to extract both form and meaning-based linguistic knowledge
from the input signal. We carry out an in-depth analysis of the representations
used by different components of the trained model and show that encoding of
semantic aspects tends to become richer as we go up the hierarchy of layers,
whereas encoding of form-related aspects of the language input tends to
initially increase and then plateau or decrease.
|
Gated Multimodal Units for Information Fusion | This paper presents a novel model for multimodal learning based on gated
neural networks. The Gated Multimodal Unit (GMU) model is intended to be used
as an internal unit in a neural network architecture whose purpose is to find
an intermediate representation based on a combination of data from different
modalities. The GMU learns to decide how modalities influence the activation of
the unit using multiplicative gates. It was evaluated on a multilabel scenario
for genre classification of movies using the plot and the poster. The GMU
improved the macro f-score performance of single-modality approaches and
outperformed other fusion strategies, including mixture of experts models.
Along with this work, the MM-IMDb dataset is released which, to the best of our
knowledge, is the largest publicly available multimodal dataset for genre
prediction on movies.
|
Truncated Variational EM for Semi-Supervised Neural Simpletrons | Inference and learning for probabilistic generative networks is often very
challenging and typically prevents scalability to as large networks as used for
deep discriminative approaches. To obtain efficiently trainable, large-scale
and well performing generative networks for semi-supervised learning, we here
combine two recent developments: a neural network reformulation of hierarchical
Poisson mixtures (Neural Simpletrons), and a novel truncated variational EM
approach (TV-EM). TV-EM provides theoretical guarantees for learning in
generative networks, and its application to Neural Simpletrons results in
particularly compact, yet approximately optimal, modifications of learning
equations. If applied to standard benchmarks, we empirically find, that
learning converges in fewer EM iterations, that the complexity per EM iteration
is reduced, and that final likelihood values are higher on average. For the
task of classification on data sets with few labels, learning improvements
result in consistently lower error rates if compared to applications without
truncation. Experiments on the MNIST data set herein allow for comparison to
standard and state-of-the-art models in the semi-supervised setting. Further
experiments on the NIST SD19 data set show the scalability of the approach when
a manifold of additional unlabeled data is available.
|
Empirical Risk Minimization for Stochastic Convex Optimization:
$O(1/n)$- and $O(1/n^2)$-type of Risk Bounds | Although there exist plentiful theories of empirical risk minimization (ERM)
for supervised learning, current theoretical understandings of ERM for a
related problem---stochastic convex optimization (SCO), are limited. In this
work, we strengthen the realm of ERM for SCO by exploiting smoothness and
strong convexity conditions to improve the risk bounds. First, we establish an
$\widetilde{O}(d/n + \sqrt{F_*/n})$ risk bound when the random function is
nonnegative, convex and smooth, and the expected function is Lipschitz
continuous, where $d$ is the dimensionality of the problem, $n$ is the number
of samples, and $F_*$ is the minimal risk. Thus, when $F_*$ is small we obtain
an $\widetilde{O}(d/n)$ risk bound, which is analogous to the
$\widetilde{O}(1/n)$ optimistic rate of ERM for supervised learning. Second, if
the objective function is also $\lambda$-strongly convex, we prove an
$\widetilde{O}(d/n + \kappa F_*/n )$ risk bound where $\kappa$ is the condition
number, and improve it to $O(1/[\lambda n^2] + \kappa F_*/n)$ when
$n=\widetilde{\Omega}(\kappa d)$. As a result, we obtain an $O(\kappa/n^2)$
risk bound under the condition that $n$ is large and $F_*$ is small, which to
the best of our knowledge, is the first $O(1/n^2)$-type of risk bound of ERM.
Third, we stress that the above results are established in a unified framework,
which allows us to derive new risk bounds under weaker conditions, e.g.,
without convexity of the random function and Lipschitz continuity of the
expected function. Finally, we demonstrate that to achieve an $O(1/[\lambda
n^2] + \kappa F_*/n)$ risk bound for supervised learning, the
$\widetilde{\Omega}(\kappa d)$ requirement on $n$ can be replaced with
$\Omega(\kappa^2)$, which is dimensionality-independent.
|
Preference-based Teaching | We introduce a new model of teaching named "preference-based teaching" and a
corresponding complexity parameter---the preference-based teaching dimension
(PBTD)---representing the worst-case number of examples needed to teach any
concept in a given concept class. Although the PBTD coincides with the
well-known recursive teaching dimension (RTD) on finite classes, it is
radically different on infinite ones: the RTD becomes infinite already for
trivial infinite classes (such as half-intervals) whereas the PBTD evaluates to
reasonably small values for a wide collection of infinite classes including
classes consisting of so-called closed sets w.r.t. a given closure operator,
including various classes related to linear sets over $\mathbb{N}_0$ (whose RTD
had been studied quite recently) and including the class of Euclidean
half-spaces. On top of presenting these concrete results, we provide the reader
with a theoretical framework (of a combinatorial flavor) which helps to derive
bounds on the PBTD.
|
Knowledge Adaptation: Teaching to Adapt | Domain adaptation is crucial in many real-world applications where the
distribution of the training data differs from the distribution of the test
data. Previous Deep Learning-based approaches to domain adaptation need to be
trained jointly on source and target domain data and are therefore unappealing
in scenarios where models need to be adapted to a large number of domains or
where a domain is evolving, e.g. spam detection where attackers continuously
change their tactics.
To fill this gap, we propose Knowledge Adaptation, an extension of Knowledge
Distillation (Bucilua et al., 2006; Hinton et al., 2015) to the domain
adaptation scenario. We show how a student model achieves state-of-the-art
results on unsupervised domain adaptation from multiple sources on a standard
sentiment analysis benchmark by taking into account the domain-specific
expertise of multiple teachers and the similarities between their domains.
When learning from a single teacher, using domain similarity to gauge
trustworthiness is inadequate. To this end, we propose a simple metric that
correlates well with the teacher's accuracy in the target domain. We
demonstrate that incorporating high-confidence examples selected by this metric
enables the student model to achieve state-of-the-art performance in the
single-source scenario.
|
Estimation of classrooms occupancy using a multi-layer perceptron | This paper presents a multi-layer perceptron model for the estimation of
classrooms number of occupants from sensed indoor environmental data-relative
humidity, air temperature, and carbon dioxide concentration. The modelling
datasets were collected from two classrooms in the Secondary School of Pombal,
Portugal. The number of occupants and occupation periods were obtained from
class attendance reports. However, post-class occupancy was unknown and the
developed model is used to reconstruct the classrooms occupancy by filling the
unreported periods. Different model structure and environment variables
combination were tested. The model with best accuracy had as input vector 10
variables of five averaged time intervals of relative humidity and carbon
dioxide concentration. The model presented a mean square error of 1.99,
coefficient of determination of 0.96 with a significance of p-value < 0.001,
and a mean absolute error of 1 occupant. These results show promising
estimation capabilities in uncertain indoor environment conditions.
|
Rapid parametric density estimation | Parametric density estimation, for example as Gaussian distribution, is the
base of the field of statistics. Machine learning requires inexpensive
estimation of much more complex densities, and the basic approach is relatively
costly maximum likelihood estimation (MLE). There will be discussed inexpensive
density estimation, for example literally fitting a polynomial (or Fourier
series) to the sample, which coefficients are calculated by just averaging
monomials (or sine/cosine) over the sample. Another discussed basic application
is fitting distortion to some standard distribution like Gaussian - analogously
to ICA, but additionally allowing to reconstruct the disturbed density.
Finally, by using weighted average, it can be also applied for estimation of
non-probabilistic densities, like modelling mass distribution, or for various
clustering problems by using negative (or complex) weights: fitting a function
which sign (or argument) determines clusters. The estimated parameters are
approaching the optimal values with error dropping like $1/\sqrt{n}$, where $n$
is the sample size.
|
Deep Learning with Dynamic Computation Graphs | Neural networks that compute over graph structures are a natural fit for
problems in a variety of domains, including natural language (parse trees) and
cheminformatics (molecular graphs). However, since the computation graph has a
different shape and size for every input, such networks do not directly support
batched training or inference. They are also difficult to implement in popular
deep learning libraries, which are based on static data-flow graphs. We
introduce a technique called dynamic batching, which not only batches together
operations between different input graphs of dissimilar shape, but also between
different nodes within a single input graph. The technique allows us to create
static graphs, using popular libraries, that emulate dynamic computation graphs
of arbitrary shape and size. We further present a high-level library of
compositional blocks that simplifies the creation of dynamic graph models.
Using the library, we demonstrate concise and batch-wise parallel
implementations for a variety of models from the literature.
|
Transfer from Multiple Linear Predictive State Representations (PSR) | In this paper, we tackle the problem of transferring policy from multiple
partially observable source environments to a partially observable target
environment modeled as predictive state representation. This is an entirely new
approach with no previous work, other than the case of transfer in fully
observable domains. We develop algorithms to successfully achieve policy
transfer when we have the model of both the source and target tasks and discuss
in detail their performance and shortcomings. These algorithms could be a
starting point for the field of transfer learning in partial observability.
|
Semi-Supervised QA with Generative Domain-Adaptive Nets | We study the problem of semi-supervised question answering----utilizing
unlabeled text to boost the performance of question answering models. We
propose a novel training framework, the Generative Domain-Adaptive Nets. In
this framework, we train a generative model to generate questions based on the
unlabeled text, and combine model-generated questions with human-generated
questions for training question answering models. We develop novel domain
adaptation algorithms, based on reinforcement learning, to alleviate the
discrepancy between the model-generated data distribution and the
human-generated data distribution. Experiments show that our proposed framework
obtains substantial improvement from unlabeled text.
|
Integration of Machine Learning Techniques to Evaluate Dynamic Customer
Segmentation Analysis for Mobile Customers | The telecommunications industry is highly competitive, which means that the
mobile providers need a business intelligence model that can be used to achieve
an optimal level of churners, as well as a minimal level of cost in marketing
activities. Machine learning applications can be used to provide guidance on
marketing strategies. Furthermore, data mining techniques can be used in the
process of customer segmentation. The purpose of this paper is to provide a
detailed analysis of the C.5 algorithm, within naive Bayesian modelling for the
task of segmenting telecommunication customers behavioural profiling according
to their billing and socio-demographic aspects. Results have been
experimentally implemented.
|
Clustering For Point Pattern Data | Clustering is one of the most common unsupervised learning tasks in machine
learning and data mining. Clustering algorithms have been used in a plethora of
applications across several scientific fields. However, there has been limited
research in the clustering of point patterns - sets or multi-sets of unordered
elements - that are found in numerous applications and data sources. In this
paper, we propose two approaches for clustering point patterns. The first is a
non-parametric method based on novel distances for sets. The second is a
model-based approach, formulated via random finite set theory, and solved by
the Expectation-Maximization algorithm. Numerical experiments show that the
proposed methods perform well on both simulated and real data.
|
Matrix Completion from $O(n)$ Samples in Linear Time | We consider the problem of reconstructing a rank-$k$ $n \times n$ matrix $M$
from a sampling of its entries. Under a certain incoherence assumption on $M$
and for the case when both the rank and the condition number of $M$ are
bounded, it was shown in \cite{CandesRecht2009, CandesTao2010, keshavan2010,
Recht2011, Jain2012, Hardt2014} that $M$ can be recovered exactly or
approximately (depending on some trade-off between accuracy and computational
complexity) using $O(n \, \text{poly}(\log n))$ samples in super-linear time
$O(n^{a} \, \text{poly}(\log n))$ for some constant $a \geq 1$.
In this paper, we propose a new matrix completion algorithm using a novel
sampling scheme based on a union of independent sparse random regular bipartite
graphs. We show that under the same conditions w.h.p. our algorithm recovers an
$\epsilon$-approximation of $M$ in terms of the Frobenius norm using $O(n
\log^2(1/\epsilon))$ samples and in linear time $O(n \log^2(1/\epsilon))$. This
provides the best known bounds both on the sample complexity and computational
complexity for reconstructing (approximately) an unknown low-rank matrix.
The novelty of our algorithm is two new steps of thresholding singular values
and rescaling singular vectors in the application of the "vanilla" alternating
minimization algorithm. The structure of sparse random regular graphs is used
heavily for controlling the impact of these regularization steps.
|
Adversarial Attacks on Neural Network Policies | Machine learning classifiers are known to be vulnerable to inputs maliciously
constructed by adversaries to force misclassification. Such adversarial
examples have been extensively studied in the context of computer vision
applications. In this work, we show adversarial attacks are also effective when
targeting neural network policies in reinforcement learning. Specifically, we
show existing adversarial example crafting techniques can be used to
significantly degrade test-time performance of trained policies. Our threat
model considers adversaries capable of introducing small perturbations to the
raw input of the policy. We characterize the degree of vulnerability across
tasks and training algorithms, for a subclass of adversarial-example attacks in
white-box and black-box settings. Regardless of the learned task or training
algorithm, we observe a significant drop in performance, even with small
adversarial perturbations that do not interfere with human perception. Videos
are available at http://rll.berkeley.edu/adversarial.
|
Data Selection Strategies for Multi-Domain Sentiment Analysis | Domain adaptation is important in sentiment analysis as sentiment-indicating
words vary between domains. Recently, multi-domain adaptation has become more
pervasive, but existing approaches train on all available source domains
including dissimilar ones. However, the selection of appropriate training data
is as important as the choice of algorithm. We undertake -- to our knowledge
for the first time -- an extensive study of domain similarity metrics in the
context of sentiment analysis and propose novel representations, metrics, and a
new scope for data selection. We evaluate the proposed methods on two
large-scale multi-domain adaptation settings on tweets and reviews and
demonstrate that they consistently outperform strong random and balanced
baselines, while our proposed selection strategy outperforms instance-level
selection and yields the best score on a large reviews corpus.
|
Trainable Greedy Decoding for Neural Machine Translation | Recent research in neural machine translation has largely focused on two
aspects; neural network architectures and end-to-end learning algorithms. The
problem of decoding, however, has received relatively little attention from the
research community. In this paper, we solely focus on the problem of decoding
given a trained neural machine translation model. Instead of trying to build a
new decoding algorithm for any specific decoding objective, we propose the idea
of trainable decoding algorithm in which we train a decoding algorithm to find
a translation that maximizes an arbitrary decoding objective. More
specifically, we design an actor that observes and manipulates the hidden state
of the neural machine translation decoder and propose to train it using a
variant of deterministic policy gradient. We extensively evaluate the proposed
algorithm using four language pairs and two decoding objectives and show that
we can indeed train a trainable greedy decoder that generates a better
translation (in terms of a target decoding objective) with minimal
computational overhead.
|
Preparing for the Unknown: Learning a Universal Policy with Online
System Identification | We present a new method of learning control policies that successfully
operate under unknown dynamic models. We create such policies by leveraging a
large number of training examples that are generated using a physical
simulator. Our system is made of two components: a Universal Policy (UP) and a
function for Online System Identification (OSI). We describe our control policy
as universal because it is trained over a wide array of dynamic models. These
variations in the dynamic model may include differences in mass and inertia of
the robots' components, variable friction coefficients, or unknown mass of an
object to be manipulated. By training the Universal Policy with this variation,
the control policy is prepared for a wider array of possible conditions when
executed in an unknown environment. The second part of our system uses the
recent state and action history of the system to predict the dynamics model
parameters mu. The value of mu from the Online System Identification is then
provided as input to the control policy (along with the system state).
Together, UP-OSI is a robust control policy that can be used across a wide
range of dynamic models, and that is also responsive to sudden changes in the
environment. We have evaluated the performance of this system on a variety of
tasks, including the problem of cart-pole swing-up, the double inverted
pendulum, locomotion of a hopper, and block-throwing of a manipulator. UP-OSI
is effective at these tasks across a wide range of dynamic models. Moreover,
when tested with dynamic models outside of the training range, UP-OSI
outperforms the Universal Policy alone, even when UP is given the actual value
of the model dynamics. In addition to the benefits of creating more robust
controllers, UP-OSI also holds out promise of narrowing the Reality Gap between
simulated and real physical systems.
|
Video Frame Synthesis using Deep Voxel Flow | We address the problem of synthesizing new video frames in an existing video,
either in-between existing frames (interpolation), or subsequent to them
(extrapolation). This problem is challenging because video appearance and
motion can be highly complex. Traditional optical-flow-based solutions often
fail where flow estimation is challenging, while newer neural-network-based
methods that hallucinate pixel values directly often produce blurry results. We
combine the advantages of these two methods by training a deep network that
learns to synthesize video frames by flowing pixel values from existing ones,
which we call deep voxel flow. Our method requires no human supervision, and
any video can be used as training data by dropping, and then learning to
predict, existing frames. The technique is efficient, and can be applied at any
video resolution. We demonstrate that our method produces results that both
quantitatively and qualitatively improve upon the state-of-the-art.
|
Deep Generalized Canonical Correlation Analysis | We present Deep Generalized Canonical Correlation Analysis (DGCCA) -- a
method for learning nonlinear transformations of arbitrarily many views of
data, such that the resulting transformations are maximally informative of each
other. While methods for nonlinear two-view representation learning (Deep CCA,
(Andrew et al., 2013)) and linear many-view representation learning
(Generalized CCA (Horst, 1961)) exist, DGCCA is the first CCA-style multiview
representation learning technique that combines the flexibility of nonlinear
(deep) representation learning with the statistical power of incorporating
information from many independent sources, or views. We present the DGCCA
formulation as well as an efficient stochastic optimization algorithm for
solving it. We learn DGCCA representations on two distinct datasets for three
downstream tasks: phonetic transcription from acoustic and articulatory
measurements, and recommending hashtags and friends on a dataset of Twitter
users. We find that DGCCA representations soundly beat existing methods at
phonetic transcription and hashtag recommendation, and in general perform no
worse than standard linear many-view techniques.
|
Deep Kernelized Autoencoders | In this paper we introduce the deep kernelized autoencoder, a neural network
model that allows an explicit approximation of (i) the mapping from an input
space to an arbitrary, user-specified kernel space and (ii) the back-projection
from such a kernel space to input space. The proposed method is based on
traditional autoencoders and is trained through a new unsupervised loss
function. During training, we optimize both the reconstruction accuracy of
input samples and the alignment between a kernel matrix given as prior and the
inner products of the hidden representations computed by the autoencoder.
Kernel alignment provides control over the hidden representation learned by the
autoencoder. Experiments have been performed to evaluate both reconstruction
and kernel alignment performance. Additionally, we applied our method to
emulate kPCA on a denoising task obtaining promising results.
|
Learning detectors of malicious web requests for intrusion detection in
network traffic | This paper proposes a generic classification system designed to detect
security threats based on the behavior of malware samples. The system relies on
statistical features computed from proxy log fields to train detectors using a
database of malware samples. The behavior detectors serve as basic reusable
building blocks of the multi-level detection architecture. The detectors
identify malicious communication exploiting encrypted URL strings and domains
generated by a Domain Generation Algorithm (DGA) which are frequently used in
Command and Control (C&C), phishing, and click fraud. Surprisingly, very
precise detectors can be built given only a limited amount of information
extracted from a single proxy log. This way, the computational requirements of
the detectors are kept low which allows for deployment on a wide range of
security devices and without depending on traffic context such as DNS logs,
Whois records, webpage content, etc. Results on several weeks of live traffic
from 100+ companies having 350k+ hosts show correct detection with a precision
exceeding 95% of malicious flows, 95% of malicious URLs and 90% of infected
hosts. In addition, a comparison with a signature and rule-based solution shows
that our system is able to detect significant amount of new threats.
|
A Modified Construction for a Support Vector Classifier to Accommodate
Class Imbalances | Given a training set with binary classification, the Support Vector Machine
identifies the hyperplane maximizing the margin between the two classes of
training data. This general formulation is useful in that it can be applied
without regard to variance differences between the classes. Ignoring these
differences is not optimal, however, as the general SVM will give the class
with lower variance an unjustifiably wide berth. This increases the chance of
misclassification of the other class and results in an overall loss of
predictive performance. An alternate construction is proposed in which the
margins of the separating hyperplane are different for each class, each
proportional to the standard deviation of its class along the direction
perpendicular to the hyperplane. The construction agrees with the SVM in the
case of equal class variances. This paper will then examine the impact to the
dual representation of the modified constraint equations.
|
Causal Regularization | In application domains such as healthcare, we want accurate predictive models
that are also causally interpretable. In pursuit of such models, we propose a
causal regularizer to steer predictive models towards causally-interpretable
solutions and theoretically study its properties. In a large-scale analysis of
Electronic Health Records (EHR), our causally-regularized model outperforms its
L1-regularized counterpart in causal accuracy and is competitive in predictive
performance. We perform non-linear causality analysis by causally regularizing
a special neural network architecture. We also show that the proposed causal
regularizer can be used together with neural representation learning algorithms
to yield up to 20% improvement over multilayer perceptron in detecting
multivariate causation, a situation common in healthcare, where many causal
factors should occur simultaneously to have an effect on the target variable.
|
Character-level Deep Conflation for Business Data Analytics | Connecting different text attributes associated with the same entity
(conflation) is important in business data analytics since it could help merge
two different tables in a database to provide a more comprehensive profile of
an entity. However, the conflation task is challenging because two text strings
that describe the same entity could be quite different from each other for
reasons such as misspelling. It is therefore critical to develop a conflation
model that is able to truly understand the semantic meaning of the strings and
match them at the semantic level. To this end, we develop a character-level
deep conflation model that encodes the input text strings from character level
into finite dimension feature vectors, which are then used to compute the
cosine similarity between the text strings. The model is trained in an
end-to-end manner using back propagation and stochastic gradient descent to
maximize the likelihood of the correct association. Specifically, we propose
two variants of the deep conflation model, based on long-short-term memory
(LSTM) recurrent neural network (RNN) and convolutional neural network (CNN),
respectively. Both models perform well on a real-world business analytics
dataset and significantly outperform the baseline bag-of-character (BoC) model.
|
EEG Representation Using Multi-instance Framework on The Manifold of
Symmetric Positive Definite Matrices for EEG-based Computer Aided Diagnosis | The generalization and robustness of an electroencephalogram (EEG)-based
computer aided diagnostic system are crucial requirements in actual clinical
practice. To reach these goals, we propose a new EEG representation that
provides a more realistic view of brain functionality by applying
multi-instance (MI) framework to consider the non-stationarity of the EEG
signal. The non-stationary characteristic of EEG is considered by describing
the signal as a bag of relevant and irrelevant concepts. The concepts are
provided by a robust representation of homogenous segments of EEG signal using
spatial covariance matrices. Due to the nonlinear geometry of the space of
covariance matrices, we determine the boundaries of the homogeneous segments
based on adaptive segmentation of the signal in a Riemannian framework. Each
subject is described as a bag of covariance matrices of homogenous segments and
the bag-level discriminative information is used for classification. To
evaluate the performance of the proposed approach, we examine it in attention
deficit hyperactivity/bipolar mood disorder detection and depression/normal
diagnosis applications. Experimental results confirm the superiority of the
proposed approach, which is gained due to the robustness of covariance
descriptor, the effectiveness of Riemannian geometry, and the benefits of
considering the inherent non-stationary nature of the brain.
|
Inductive Pairwise Ranking: Going Beyond the n log(n) Barrier | We study the problem of ranking a set of items from nonactively chosen
pairwise preferences where each item has feature information with it. We
propose and characterize a very broad class of preference matrices giving rise
to the Feature Low Rank (FLR) model, which subsumes several models ranging from
the classic Bradley-Terry-Luce (BTL) (Bradley and Terry 1952) and Thurstone
(Thurstone 1927) models to the recently proposed blade-chest (Chen and Joachims
2016) and generic low-rank preference (Rajkumar and Agarwal 2016) models. We
use the technique of matrix completion in the presence of side information to
develop the Inductive Pairwise Ranking (IPR) algorithm that provably learns a
good ranking under the FLR model, in a sample-efficient manner. In practice,
through systematic synthetic simulations, we confirm our theoretical findings
regarding improvements in the sample complexity due to the use of feature
information. Moreover, on popular real-world preference learning datasets, with
as less as 10% sampling of the pairwise comparisons, our method recovers a good
ranking.
|
Energy Saving Additive Neural Network | In recent years, machine learning techniques based on neural networks for
mobile computing become increasingly popular. Classical multi-layer neural
networks require matrix multiplications at each stage. Multiplication operation
is not an energy efficient operation and consequently it drains the battery of
the mobile device. In this paper, we propose a new energy efficient neural
network with the universal approximation property over space of Lebesgue
integrable functions. This network, called, additive neural network, is very
suitable for mobile computing. The neural structure is based on a novel vector
product definition, called ef-operator, that permits a multiplier-free
implementation. In ef-operation, the "product" of two real numbers is defined
as the sum of their absolute values, with the sign determined by the sign of
the product of the numbers. This "product" is used to construct a vector
product in $R^N$. The vector product induces the $l_1$ norm. The proposed
additive neural network successfully solves the XOR problem. The experiments on
MNIST dataset show that the classification performances of the proposed
additive neural networks are very similar to the corresponding multi-layer
perceptron and convolutional neural networks (LeNet).
|
Rate Optimal Estimation and Confidence Intervals for High-dimensional
Regression with Missing Covariates | Although a majority of the theoretical literature in high-dimensional
statistics has focused on settings which involve fully-observed data, settings
with missing values and corruptions are common in practice. We consider the
problems of estimation and of constructing component-wise confidence intervals
in a sparse high-dimensional linear regression model when some covariates of
the design matrix are missing completely at random. We analyze a variant of the
Dantzig selector [9] for estimating the regression model and we use a
de-biasing argument to construct component-wise confidence intervals. Our first
main result is to establish upper bounds on the estimation error as a function
of the model parameters (the sparsity level s, the expected fraction of
observed covariates $\rho_*$, and a measure of the signal strength
$\|\beta^*\|_2$). We find that even in an idealized setting where the
covariates are assumed to be missing completely at random, somewhat
surprisingly and in contrast to the fully-observed setting, there is a
dichotomy in the dependence on model parameters and much faster rates are
obtained if the covariance matrix of the random design is known. To study this
issue further, our second main contribution is to provide lower bounds on the
estimation error showing that this discrepancy in rates is unavoidable in a
minimax sense. We then consider the problem of high-dimensional inference in
the presence of missing data. We construct and analyze confidence intervals
using a de-biased estimator. In the presence of missing data, inference is
complicated by the fact that the de-biasing matrix is correlated with the pilot
estimator and this necessitates the design of a new estimator and a novel
analysis. We also complement our mathematical study with extensive simulations
on synthetic and semi-synthetic data that show the accuracy of our asymptotic
predictions for finite sample sizes.
|
A Fast and Scalable Joint Estimator for Learning Multiple Related Sparse
Gaussian Graphical Models | Estimating multiple sparse Gaussian Graphical Models (sGGMs) jointly for many
related tasks (large $K$) under a high-dimensional (large $p$) situation is an
important task. Most previous studies for the joint estimation of multiple
sGGMs rely on penalized log-likelihood estimators that involve expensive and
difficult non-smooth optimizations. We propose a novel approach, FASJEM for
\underline{fa}st and \underline{s}calable \underline{j}oint
structure-\underline{e}stimation of \underline{m}ultiple sGGMs at a large
scale. As the first study of joint sGGM using the Elementary Estimator
framework, our work has three major contributions: (1) We solve FASJEM through
an entry-wise manner which is parallelizable. (2) We choose a proximal
algorithm to optimize FASJEM. This improves the computational efficiency from
$O(Kp^3)$ to $O(Kp^2)$ and reduces the memory requirement from $O(Kp^2)$ to
$O(K)$. (3) We theoretically prove that FASJEM achieves a consistent estimation
with a convergence rate of $O(\log(Kp)/n_{tot})$. On several synthetic and four
real-world datasets, FASJEM shows significant improvements over baselines on
accuracy, computational complexity, and memory costs.
|
Joint Discovery of Object States and Manipulation Actions | Many human activities involve object manipulations aiming to modify the
object state. Examples of common state changes include full/empty bottle,
open/closed door, and attached/detached car wheel. In this work, we seek to
automatically discover the states of objects and the associated manipulation
actions. Given a set of videos for a particular task, we propose a joint model
that learns to identify object states and to localize state-modifying actions.
Our model is formulated as a discriminative clustering cost with constraints.
We assume a consistent temporal order for the changes in object states and
manipulation actions, and introduce new optimization techniques to learn model
parameters without additional supervision. We demonstrate successful discovery
of seven manipulation actions and corresponding object states on a new dataset
of videos depicting real-life object manipulations. We show that our joint
formulation results in an improvement of object state discovery by action
recognition and vice versa.
|
Graph Based Relational Features for Collective Classification | Statistical Relational Learning (SRL) methods have shown that classification
accuracy can be improved by integrating relations between samples. Techniques
such as iterative classification or relaxation labeling achieve this by
propagating information between related samples during the inference process.
When only a few samples are labeled and connections between samples are sparse,
collective inference methods have shown large improvements over standard
feature-based ML methods. However, in contrast to feature based ML, collective
inference methods require complex inference procedures and often depend on the
strong assumption of label consistency among related samples. In this paper, we
introduce new relational features for standard ML methods by extracting
information from direct and indirect relations. We show empirically on three
standard benchmark datasets that our relational features yield results
comparable to collective inference methods. Finally we show that our proposal
outperforms these methods when additional information is available.
|
Minimax Lower Bounds for Ridge Combinations Including Neural Nets | Estimation of functions of $ d $ variables is considered using ridge
combinations of the form $ \textstyle\sum_{k=1}^m c_{1,k}
\phi(\textstyle\sum_{j=1}^d c_{0,j,k}x_j-b_k) $ where the activation function $
\phi $ is a function with bounded value and derivative. These include
single-hidden layer neural networks, polynomials, and sinusoidal models. From a
sample of size $ n $ of possibly noisy values at random sites $ X \in B =
[-1,1]^d $, the minimax mean square error is examined for functions in the
closure of the $ \ell_1 $ hull of ridge functions with activation $ \phi $. It
is shown to be of order $ d/n $ to a fractional power (when $ d $ is of smaller
order than $ n $), and to be of order $ (\log d)/n $ to a fractional power
(when $ d $ is of larger order than $ n $). Dependence on constraints $ v_0 $
and $ v_1 $ on the $ \ell_1 $ norms of inner parameter $ c_0 $ and outer
parameter $ c_1 $, respectively, is also examined. Also, lower and upper bounds
on the fractional power are given. The heart of the analysis is development of
information-theoretic packing numbers for these classes of functions.
|
Coordinated Online Learning With Applications to Learning User
Preferences | We study an online multi-task learning setting, in which instances of related
tasks arrive sequentially, and are handled by task-specific online learners. We
consider an algorithmic framework to model the relationship of these tasks via
a set of convex constraints. To exploit this relationship, we design a novel
algorithm -- COOL -- for coordinating the individual online learners: Our key
idea is to coordinate their parameters via weighted projections onto a convex
set. By adjusting the rate and accuracy of the projection, the COOL algorithm
allows for a trade-off between the benefit of coordination and the required
computation/communication. We derive regret bounds for our approach and analyze
how they are influenced by these trade-off factors. We apply our results on the
application of learning users' preferences on the Airbnb marketplace with the
goal of incentivizing users to explore under-reviewed apartments.
|
Multi-feature classifiers for burst detection in single EEG channels
from preterm infants | The study of electroencephalographic (EEG) bursts in preterm infants provides
valuable information about maturation or prognostication after perinatal
asphyxia. Over the last two decades, a number of works proposed algorithms to
automatically detect EEG bursts in preterm infants, but they were designed for
populations under 35 weeks of post menstrual age (PMA). However, as the brain
activity evolves rapidly during postnatal life, these solutions might be
under-performing with increasing PMA. In this work we focused on preterm
infants reaching term ages (PMA $\geq$ 36 weeks) using multi-feature
classification on a single EEG channel. Five EEG burst detectors relying on
different machine learning approaches were compared: Logistic regression (LR),
linear discriminant analysis (LDA), k-nearest neighbors (kNN), support vector
machines (SVM) and thresholding (Th). Classifiers were trained by visually
labeled EEG recordings from 14 very preterm infants (born after 28 weeks of
gestation) with 36 - 41 weeks PMA. The most performing classifiers reached
about 95\% accuracy (kNN, SVM and LR) whereas Th obtained 84\%. Compared to
human-automatic agreements, LR provided the highest scores (Cohen's kappa =
0.71) and the best computational efficiency using only three EEG features.
Applying this classifier in a test database of 21 infants $\geq$ 36 weeks PMA,
we show that long EEG bursts and short inter-bust periods are characteristic of
infants with the highest PMA and weights. In view of these results, LR-based
burst detection could be a suitable tool to study maturation in monitoring or
portable devices using a single EEG channel.
|
Policy Learning with Observational Data | In many areas, practitioners seek to use observational data to learn a
treatment assignment policy that satisfies application-specific constraints,
such as budget, fairness, simplicity, or other functional form constraints. For
example, policies may be restricted to take the form of decision trees based on
a limited set of easily observable individual characteristics. We propose a new
approach to this problem motivated by the theory of semiparametrically
efficient estimation. Our method can be used to optimize either binary
treatments or infinitesimal nudges to continuous treatments, and can leverage
observational data where causal effects are identified using a variety of
strategies, including selection on observables and instrumental variables.
Given a doubly robust estimator of the causal effect of assigning everyone to
treatment, we develop an algorithm for choosing whom to treat, and establish
strong guarantees for the asymptotic utilitarian regret of the resulting
policy.
|
Online and Offline Domain Adaptation for Reducing BCI Calibration Effort | Many real-world brain-computer interface (BCI) applications rely on
single-trial classification of event-related potentials (ERPs) in EEG signals.
However, because different subjects have different neural responses to even the
same stimulus, it is very difficult to build a generic ERP classifier whose
parameters fit all subjects. The classifier needs to be calibrated for each
individual subject, using some labeled subject-specific data. This paper
proposes both online and offline weighted adaptation regularization (wAR)
algorithms to reduce this calibration effort, i.e., to minimize the amount of
labeled subject-specific EEG data required in BCI calibration, and hence to
increase the utility of the BCI system. We demonstrate using a visually-evoked
potential oddball task and three different EEG headsets that both online and
offline wAR algorithms significantly outperform several other algorithms.
Moreover, through source domain selection, we can reduce their computational
cost by about 50%, making them more suitable for real-time applications.
|
Driver Drowsiness Estimation from EEG Signals Using Online Weighted
Adaptation Regularization for Regression (OwARR) | One big challenge that hinders the transition of brain-computer interfaces
(BCIs) from laboratory settings to real-life applications is the availability
of high-performance and robust learning algorithms that can effectively handle
individual differences, i.e., algorithms that can be applied to a new subject
with zero or very little subject-specific calibration data. Transfer learning
and domain adaptation have been extensively used for this purpose. However,
most previous works focused on classification problems. This paper considers an
important regression problem in BCI, namely, online driver drowsiness
estimation from EEG signals. By integrating fuzzy sets with domain adaptation,
we propose a novel online weighted adaptation regularization for regression
(OwARR) algorithm to reduce the amount of subject-specific calibration data,
and also a source domain selection (SDS) approach to save about half of the
computational cost of OwARR. Using a simulated driving dataset with 15
subjects, we show that OwARR and OwARR-SDS can achieve significantly smaller
estimation errors than several other approaches. We also provide comprehensive
analyses on the robustness of OwARR and OwARR-SDS.
|
Switching EEG Headsets Made Easy: Reducing Offline Calibration Effort
Using Active Weighted Adaptation Regularization | Electroencephalography (EEG) headsets are the most commonly used sensing
devices for Brain-Computer Interface. In real-world applications, there are
advantages to extrapolating data from one user session to another. However,
these advantages are limited if the data arise from different hardware systems,
which often vary between application spaces. Currently, this creates a need to
recalibrate classifiers, which negatively affects people's interest in using
such systems. In this paper, we employ active weighted adaptation
regularization (AwAR), which integrates weighted adaptation regularization
(wAR) and active learning, to expedite the calibration process. wAR makes use
of labeled data from the previous headset and handles class-imbalance, and
active learning selects the most informative samples from the new headset to
label. Experiments on single-trial event-related potential classification show
that AwAR can significantly increase the classification accuracy, given the
same number of labeled samples from the new headset. In other words, AwAR can
effectively reduce the number of labeled samples required from the new headset,
given a desired classification accuracy, suggesting value in collating data for
use in wide scale transfer-learning applications.
|
Spatial Filtering for EEG-Based Regression Problems in Brain-Computer
Interface (BCI) | Electroencephalogram (EEG) signals are frequently used in brain-computer
interfaces (BCIs), but they are easily contaminated by artifacts and noises, so
preprocessing must be done before they are fed into a machine learning
algorithm for classification or regression. Spatial filters have been widely
used to increase the signal-to-noise ratio of EEG for BCI classification
problems, but their applications in BCI regression problems have been very
limited. This paper proposes two common spatial pattern (CSP) filters for
EEG-based regression problems in BCI, which are extended from the CSP filter
for classification, by making use of fuzzy sets. Experimental results on
EEG-based response speed estimation from a large-scale study, which collected
143 sessions of sustained-attention psychomotor vigilance task data from 17
subjects during a 5-month period, demonstrate that the two proposed spatial
filters can significantly increase the EEG signal quality. When used in LASSO
and k-nearest neighbors regression for user response speed estimation, the
spatial filters can reduce the root mean square estimation error by
10.02-19.77%, and at the same time increase the correlation to the true
response speed by 19.39-86.47%.
|
Fixing an error in Caponnetto and de Vito (2007) | The seminal paper of Caponnetto and de Vito (2007) provides minimax-optimal
rates for kernel ridge regression in a very general setting. Its proof,
however, contains an error in its bound on the effective dimensionality. In
this note, we explain the mistake, provide a correct bound, and show that the
main theorem remains true.
|
Multi-step Off-policy Learning Without Importance Sampling Ratios | To estimate the value functions of policies from exploratory data, most
model-free off-policy algorithms rely on importance sampling, where the use of
importance sampling ratios often leads to estimates with severe variance. It is
thus desirable to learn off-policy without using the ratios. However, such an
algorithm does not exist for multi-step learning with function approximation.
In this paper, we introduce the first such algorithm based on
temporal-difference (TD) learning updates. We show that an explicit use of
importance sampling ratios can be eliminated by varying the amount of
bootstrapping in TD updates in an action-dependent manner. Our new algorithm
achieves stability using a two-timescale gradient-based TD update. A prior
algorithm based on lookup table representation called Tree Backup can also be
retrieved using action-dependent bootstrapping, becoming a special case of our
algorithm. In two challenging off-policy tasks, we demonstrate that our
algorithm is stable, effectively avoids the large variance issue, and can
perform substantially better than its state-of-the-art counterpart.
|
Multi-agent Reinforcement Learning in Sequential Social Dilemmas | Matrix games like Prisoner's Dilemma have guided research on social dilemmas
for decades. However, they necessarily treat the choice to cooperate or defect
as an atomic action. In real-world social dilemmas these choices are temporally
extended. Cooperativeness is a property that applies to policies, not
elementary actions. We introduce sequential social dilemmas that share the
mixed incentive structure of matrix game social dilemmas but also require
agents to learn policies that implement their strategic intentions. We analyze
the dynamics of policies learned by multiple self-interested independent
learning agents, each using its own deep Q-network, on two Markov games we
introduce here: 1. a fruit Gathering game and 2. a Wolfpack hunting game. We
characterize how learned behavior in each domain changes as a function of
environmental factors including resource abundance. Our experiments show how
conflict can emerge from competition over shared resources and shed light on
how the sequential nature of real world social dilemmas affects cooperation.
|
Following the Leader and Fast Rates in Linear Prediction: Curved
Constraint Sets and Other Regularities | The follow the leader (FTL) algorithm, perhaps the simplest of all online
learning algorithms, is known to perform well when the loss functions it is
used on are convex and positively curved. In this paper we ask whether there
are other "lucky" settings when FTL achieves sublinear, "small" regret. In
particular, we study the fundamental problem of linear prediction over a
non-empty convex, compact domain. Amongst other results, we prove that the
curvature of the boundary of the domain can act as if the losses were curved:
In this case, we prove that as long as the mean of the loss vectors have
positive lengths bounded away from zero, FTL enjoys a logarithmic growth rate
of regret, while, e.g., for polytope domains and stochastic data it enjoys
finite expected regret. Building on a previously known meta-algorithm, we also
get an algorithm that simultaneously enjoys the worst-case guarantees and the
bound available for FTL.
|
Sigmoid-Weighted Linear Units for Neural Network Function Approximation
in Reinforcement Learning | In recent years, neural networks have enjoyed a renaissance as function
approximators in reinforcement learning. Two decades after Tesauro's TD-Gammon
achieved near top-level human performance in backgammon, the deep reinforcement
learning algorithm DQN achieved human-level performance in many Atari 2600
games. The purpose of this study is twofold. First, we propose two activation
functions for neural network function approximation in reinforcement learning:
the sigmoid-weighted linear unit (SiLU) and its derivative function (dSiLU).
The activation of the SiLU is computed by the sigmoid function multiplied by
its input. Second, we suggest that the more traditional approach of using
on-policy learning with eligibility traces, instead of experience replay, and
softmax action selection with simple annealing can be competitive with DQN,
without the need for a separate target network. We validate our proposed
approach by, first, achieving new state-of-the-art results in both stochastic
SZ-Tetris and Tetris with a small 10$\times$10 board, using TD($\lambda$)
learning and shallow dSiLU network agents, and, then, by outperforming DQN in
the Atari 2600 domain by using a deep Sarsa($\lambda$) agent with SiLU and
dSiLU hidden units.
|
Supervised Learning Based Algorithm Selection for Deep Neural Networks | Many recent deep learning platforms rely on third-party libraries (such as
cuBLAS) to utilize the computing power of modern hardware accelerators (such as
GPUs). However, we observe that they may achieve suboptimal performance because
the library functions are not used appropriately. In this paper, we target at
optimizing the operations of multiplying a matrix with the transpose of another
matrix (referred to as NT operation hereafter), which contribute about half of
the training time of fully connected deep neural networks. Rather than directly
calling the library function, we propose a supervised learning based algorithm
selection approach named MTNN, which uses a gradient boosted decision tree to
select one from two alternative NT implementations intelligently: (1) calling
the cuBLAS library function; (2) calling our proposed algorithm TNN that uses
an efficient out-of-place matrix transpose. We evaluate the performance of MTNN
on two modern GPUs: NVIDIA GTX 1080 and NVIDIA Titan X Pascal. MTNN can achieve
96\% of prediction accuracy with very low computational overhead, which results
in an average of 54\% performance improvement on a range of NT operations. To
further evaluate the impact of MTNN on the training process of deep neural
networks, we have integrated MTNN into a popular deep learning platform Caffe.
Our experimental results show that the revised Caffe can outperform the
original one by an average of 28\%. Both MTNN and the revised Caffe are
open-source.
|
Adaptive and Resilient Soft Tensegrity Robots | Living organisms intertwine soft (e.g., muscle) and hard (e.g., bones)
materials, giving them an intrinsic flexibility and resiliency often lacking in
conventional rigid robots. The emerging field of soft robotics seeks to harness
these same properties in order to create resilient machines. The nature of soft
materials, however, presents considerable challenges to aspects of design,
construction, and control -- and up until now, the vast majority of gaits for
soft robots have been hand-designed through empirical trial-and-error. This
manuscript describes an easy-to-assemble tensegrity-based soft robot capable of
highly dynamic locomotive gaits and demonstrating structural and behavioral
resilience in the face of physical damage. Enabling this is the use of a
machine learning algorithm able to discover effective gaits with a minimal
number of physical trials. These results lend further credence to soft-robotic
approaches that seek to harness the interaction of complex material dynamics in
order to generate a wealth of dynamical behaviors.
|
A Deterministic and Generalized Framework for Unsupervised Learning with
Restricted Boltzmann Machines | Restricted Boltzmann machines (RBMs) are energy-based neural-networks which
are commonly used as the building blocks for deep architectures neural
architectures. In this work, we derive a deterministic framework for the
training, evaluation, and use of RBMs based upon the Thouless-Anderson-Palmer
(TAP) mean-field approximation of widely-connected systems with weak
interactions coming from spin-glass theory. While the TAP approach has been
extensively studied for fully-visible binary spin systems, our construction is
generalized to latent-variable models, as well as to arbitrarily distributed
real-valued spin systems with bounded support. In our numerical experiments, we
demonstrate the effective deterministic training of our proposed models and are
able to show interesting features of unsupervised learning which could not be
directly observed with sampling. Additionally, we demonstrate how to utilize
our TAP-based framework for leveraging trained RBMs as joint priors in
denoising problems.
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.