title
stringlengths 7
246
| abstract
stringlengths 6
3.31k
|
|---|---|
A Mean Field View of the Landscape of Two-Layers Neural Networks | Multi-layer neural networks are among the most powerful models in machine
learning, yet the fundamental reasons for this success defy mathematical
understanding. Learning a neural network requires to optimize a non-convex
high-dimensional objective (risk function), a problem which is usually attacked
using stochastic gradient descent (SGD). Does SGD converge to a global optimum
of the risk or only to a local optimum? In the first case, does this happen
because local minima are absent, or because SGD somehow avoids them? In the
second, why do local minima reached by SGD have good generalization properties?
In this paper we consider a simple case, namely two-layers neural networks,
and prove that -in a suitable scaling limit- SGD dynamics is captured by a
certain non-linear partial differential equation (PDE) that we call
distributional dynamics (DD). We then consider several specific examples, and
show how DD can be used to prove convergence of SGD to networks with nearly
ideal generalization error. This description allows to 'average-out' some of
the complexities of the landscape of neural networks, and can be used to prove
a general convergence result for noisy SGD.
|
Visualizing the Feature Importance for Black Box Models | In recent years, a large amount of model-agnostic methods to improve the
transparency, trustability and interpretability of machine learning models have
been developed. We introduce local feature importance as a local version of a
recent model-agnostic global feature importance method. Based on local feature
importance, we propose two visual tools: partial importance (PI) and individual
conditional importance (ICI) plots which visualize how changes in a feature
affect the model performance on average, as well as for individual
observations. Our proposed methods are related to partial dependence (PD) and
individual conditional expectation (ICE) plots, but visualize the expected
(conditional) feature importance instead of the expected (conditional)
prediction. Furthermore, we show that averaging ICI curves across observations
yields a PI curve, and integrating the PI curve with respect to the
distribution of the considered feature results in the global feature
importance. Another contribution of our paper is the Shapley feature
importance, which fairly distributes the overall performance of a model among
the features according to the marginal contributions and which can be used to
compare the feature importance across different models.
|
Bayesian Metabolic Flux Analysis reveals intracellular flux couplings | Metabolic flux balance analyses are a standard tool in analysing metabolic
reaction rates compatible with measurements, steady-state and the metabolic
reaction network stoichiometry. Flux analysis methods commonly place
unrealistic assumptions on fluxes due to the convenience of formulating the
problem as a linear programming model, and most methods ignore the notable
uncertainty in flux estimates. We introduce a novel paradigm of Bayesian
metabolic flux analysis that models the reactions of the whole genome-scale
cellular system in probabilistic terms, and can infer the full flux vector
distribution of genome-scale metabolic systems based on exchange and
intracellular (e.g. 13C) flux measurements, steady-state assumptions, and
target function assumptions. The Bayesian model couples all fluxes jointly
together in a simple truncated multivariate posterior distribution, which
reveals informative flux couplings. Our model is a plug-in replacement to
conventional metabolic balance methods, such as flux balance analysis (FBA).
Our experiments indicate that we can characterise the genome-scale flux
covariances, reveal flux couplings, and determine more intracellular unobserved
fluxes in C. acetobutylicum from 13C data than flux variability analysis. The
COBRA compatible software is available at github.com/markusheinonen/bamfa
|
Understanding Neural Networks and Individual Neuron Importance via
Information-Ordered Cumulative Ablation | In this work, we investigate the use of three information-theoretic
quantities -- entropy, mutual information with the class variable, and a class
selectivity measure based on Kullback-Leibler divergence -- to understand and
study the behavior of already trained fully-connected feed-forward neural
networks. We analyze the connection between these information-theoretic
quantities and classification performance on the test set by cumulatively
ablating neurons in networks trained on MNIST, FashionMNIST, and CIFAR-10. Our
results parallel those recently published by Morcos et al., indicating that
class selectivity is not a good indicator for classification performance.
However, looking at individual layers separately, both mutual information and
class selectivity are positively correlated with classification performance, at
least for networks with ReLU activation functions. We provide explanations for
this phenomenon and conclude that it is ill-advised to compare the proposed
information-theoretic quantities across layers. Furthermore, we show that
cumulative ablation of neurons with ascending or descending
information-theoretic quantities can be used to formulate hypotheses regarding
the joint behavior of multiple neurons, such as redundancy and synergy, with
comparably low computational cost. We also draw connections to the information
bottleneck theory for neural networks.
|
Are ResNets Provably Better than Linear Predictors? | A residual network (or ResNet) is a standard deep neural net architecture,
with state-of-the-art performance across numerous applications. The main
premise of ResNets is that they allow the training of each layer to focus on
fitting just the residual of the previous layer's output and the target output.
Thus, we should expect that the trained network is no worse than what we can
obtain if we remove the residual layers and train a shallower network instead.
However, due to the non-convexity of the optimization problem, it is not at all
clear that ResNets indeed achieve this behavior, rather than getting stuck at
some arbitrarily poor local minimum. In this paper, we rigorously prove that
arbitrarily deep, nonlinear residual units indeed exhibit this behavior, in the
sense that the optimization landscape contains no local minima with value above
what can be obtained with a linear predictor (namely a 1-layer network).
Notably, we show this under minimal or no assumptions on the precise network
architecture, data distribution, or loss function used. We also provide a
quantitative analysis of approximate stationary points for this problem.
Finally, we show that with a certain tweak to the architecture, training the
network with standard stochastic gradient descent achieves an objective value
close or better than any linear predictor.
|
Exact Distributed Training: Random Forest with Billions of Examples | We introduce an exact distributed algorithm to train Random Forest models as
well as other decision forest models without relying on approximating best
split search. We explain the proposed algorithm and compare it to related
approaches for various complexity measures (time, ram, disk, and network
complexity analysis). We report its running performances on artificial and
real-world datasets of up to 18 billions examples. This figure is several
orders of magnitude larger than datasets tackled in the existing literature.
Finally, we empirically show that Random Forest benefits from being trained on
more data, even in the case of already gigantic datasets. Given a dataset with
17.3B examples with 82 features (3 numerical, other categorical with high
arity), our implementation trains a tree in 22h.
|
CoNet: Collaborative Cross Networks for Cross-Domain Recommendation | The cross-domain recommendation technique is an effective way of alleviating
the data sparse issue in recommender systems by leveraging the knowledge from
relevant domains. Transfer learning is a class of algorithms underlying these
techniques. In this paper, we propose a novel transfer learning approach for
cross-domain recommendation by using neural networks as the base model. In
contrast to the matrix factorization based cross-domain techniques, our method
is deep transfer learning, which can learn complex user-item interaction
relationships. We assume that hidden layers in two base networks are connected
by cross mappings, leading to the collaborative cross networks (CoNet). CoNet
enables dual knowledge transfer across domains by introducing cross connections
from one base network to another and vice versa. CoNet is achieved in
multi-layer feedforward networks by adding dual connections and joint loss
functions, which can be trained efficiently by back-propagation. The proposed
model is thoroughly evaluated on two large real-world datasets. It outperforms
baselines by relative improvements of 7.84\% in NDCG. We demonstrate the
necessity of adaptively selecting representations to transfer. Our model can
reduce tens of thousands training examples comparing with non-transfer methods
and still has the competitive performance with them.
|
Improving Long-Horizon Forecasts with Expectation-Biased LSTM Networks | State-of-the-art forecasting methods using Recurrent Neural Net- works (RNN)
based on Long-Short Term Memory (LSTM) cells have shown exceptional performance
targeting short-horizon forecasts, e.g given a set of predictor features,
forecast a target value for the next few time steps in the future. However, in
many applica- tions, the performance of these methods decays as the forecasting
horizon extends beyond these few time steps. This paper aims to explore the
challenges of long-horizon forecasting using LSTM networks. Here, we illustrate
the long-horizon forecasting problem in datasets from neuroscience and energy
supply management. We then propose expectation-biasing, an approach motivated
by the literature of Dynamic Belief Networks, as a solution to improve
long-horizon forecasting using LSTMs. We propose two LSTM ar- chitectures along
with two methods for expectation biasing that significantly outperforms
standard practice.
|
Entropic Spectral Learning for Large-Scale Graphs | Graph spectra have been successfully used to classify network types, compute
the similarity between graphs, and determine the number of communities in a
network. For large graphs, where an eigen-decomposition is infeasible,
iterative moment matched approximations to the spectra and kernel smoothing are
typically used. We show that the underlying moment information is lost when
using kernel smoothing. We further propose a spectral density approximation
based on the method of Maximum Entropy, for which we develop a new algorithm.
This method matches moments exactly and is everywhere positive. We demonstrate
its effectiveness and superiority over existing approaches in learning graph
spectra, via experiments on both synthetic networks, such as the
Erd\H{o}s-R\'{e}nyi and Barab\'{a}si-Albert random graphs, and real-world
networks, such as the social networks for Orkut, YouTube, and Amazon from the
SNAP dataset.
|
Active choice of teachers, learning strategies and goals for a socially
guided intrinsic motivation learner | We present an active learning architecture that allows a robot to actively
learn which data collection strategy is most efficient for acquiring motor
skills to achieve multiple outcomes, and generalise over its experience to
achieve new outcomes. The robot explores its environment both via interactive
learning and goal-babbling. It learns at the same time when, who and what to
actively imitate from several available teachers, and learns when not to use
social guidance but use active goal-oriented self-exploration. This is
formalised in the framework of life-long strategic learning. The proposed
architecture, called Socially Guided Intrinsic Motivation with Active Choice of
Teacher and Strategy (SGIM-ACTS), relies on hierarchical active decisions of
what and how to learn driven by empirical evaluation of learning progress for
each learning strategy. We illustrate with an experiment where a simulated
robot learns to control its arm for realising two kinds of different outcomes.
It has to choose actively and hierarchically at each learning episode: 1) what
to learn: which outcome is most interesting to select as a goal to focus on for
goal-directed exploration; 2) how to learn: which data collection strategy to
use among self-exploration, mimicry and emulation; 3) once he has decided when
and what to imitate by choosing mimicry or emulation, then he has to choose who
to imitate, from a set of different teachers. We show that SGIM-ACTS learns
significantly more efficiently than using single learning strategies, and
coherently selects the best strategy with respect to the chosen outcome, taking
advantage of the available teachers (with different levels of skills).
|
Co-teaching: Robust Training of Deep Neural Networks with Extremely
Noisy Labels | Deep learning with noisy labels is practically challenging, as the capacity
of deep models is so high that they can totally memorize these noisy labels
sooner or later during training. Nonetheless, recent studies on the
memorization effects of deep neural networks show that they would first
memorize training data of clean labels and then those of noisy labels.
Therefore in this paper, we propose a new deep learning paradigm called
Co-teaching for combating with noisy labels. Namely, we train two deep neural
networks simultaneously, and let them teach each other given every mini-batch:
firstly, each network feeds forward all data and selects some data of possibly
clean labels; secondly, two networks communicate with each other what data in
this mini-batch should be used for training; finally, each network back
propagates the data selected by its peer network and updates itself. Empirical
results on noisy versions of MNIST, CIFAR-10 and CIFAR-100 demonstrate that
Co-teaching is much superior to the state-of-the-art methods in the robustness
of trained deep models.
|
A Study on Overfitting in Deep Reinforcement Learning | Recent years have witnessed significant progresses in deep Reinforcement
Learning (RL). Empowered with large scale neural networks, carefully designed
architectures, novel training algorithms and massively parallel computing
devices, researchers are able to attack many challenging RL problems. However,
in machine learning, more training power comes with a potential risk of more
overfitting. As deep RL techniques are being applied to critical problems such
as healthcare and finance, it is important to understand the generalization
behaviors of the trained agents. In this paper, we conduct a systematic study
of standard RL agents and find that they could overfit in various ways.
Moreover, overfitting could happen "robustly": commonly used techniques in RL
that add stochasticity do not necessarily prevent or detect overfitting. In
particular, the same agents and learning algorithms could have drastically
different test performance, even when all of them achieve optimal rewards
during training. The observations call for more principled and careful
evaluation protocols in RL. We conclude with a general discussion on
overfitting in RL and a study of the generalization behaviors from the
perspective of inductive bias.
|
A Multi-task Selected Learning Approach for Solving 3D Flexible Bin
Packing Problem | A 3D flexible bin packing problem (3D-FBPP) arises from the process of
warehouse packing in e-commerce. An online customer's order usually contains
several items and needs to be packed as a whole before shipping. In particular,
5% of tens of millions of packages are using plastic wrapping as outer
packaging every day, which brings pressure on the plastic surface minimization
to save traditional logistics costs. Because of the huge practical
significance, we focus on the issue of packing cuboid-shaped items orthogonally
into a least-surface-area bin. The existing heuristic methods for classic 3D
bin packing don't work well for this particular NP-hard problem and designing a
good problem-specific heuristic is non-trivial. In this paper, rather than
designing heuristics, we propose a novel multi-task framework based on Selected
Learning to learn a heuristic-like policy that generates the sequence and
orientations of items to be packed simultaneously. Through comprehensive
experiments on a large scale real-world transaction order dataset and online AB
tests, we show: 1) our selected learning method trades off the imbalance and
correlation among the tasks and significantly outperforms the single task
Pointer Network and the multi-task network without selected learning; 2) our
method obtains an average 5.47% cost reduction than the well-designed greedy
algorithm which is previously used in our online production system.
|
Modeling and Simultaneously Removing Bias via Adversarial Neural
Networks | In real world systems, the predictions of deployed Machine Learned models
affect the training data available to build subsequent models. This introduces
a bias in the training data that needs to be addressed. Existing solutions to
this problem attempt to resolve the problem by either casting this in the
reinforcement learning framework or by quantifying the bias and re-weighting
the loss functions. In this work, we develop a novel Adversarial Neural Network
(ANN) model, an alternative approach which creates a representation of the data
that is invariant to the bias. We take the Paid Search auction as our working
example and ad display position features as the confounding features for this
setting. We show the success of this approach empirically on both synthetic
data as well as real world paid search auction data from a major search engine.
|
K-Nearest Oracles Borderline Dynamic Classifier Ensemble Selection | Dynamic Ensemble Selection (DES) techniques aim to select locally competent
classifiers for the classification of each new test sample. Most DES techniques
estimate the competence of classifiers using a given criterion over the region
of competence of the test sample (its the nearest neighbors in the validation
set). The K-Nearest Oracles Eliminate (KNORA-E) DES selects all classifiers
that correctly classify all samples in the region of competence of the test
sample, if such classifier exists, otherwise, it removes from the region of
competence the sample that is furthest from the test sample, and the process
repeats. When the region of competence has samples of different classes,
KNORA-E can reduce the region of competence in such a way that only samples of
a single class remain in the region of competence, leading to the selection of
locally incompetent classifiers that classify all samples in the region of
competence as being from the same class. In this paper, we propose two DES
techniques: K-Nearest Oracles Borderline (KNORA-B) and K-Nearest Oracles
Borderline Imbalanced (KNORA-BI). KNORA-B is a DES technique based on KNORA-E
that reduces the region of competence but maintains at least one sample from
each class that is in the original region of competence. KNORA-BI is a
variation of KNORA-B for imbalance datasets that reduces the region of
competence but maintains at least one minority class sample if there is any in
the original region of competence. Experiments are conducted comparing the
proposed techniques with 19 DES techniques from the literature using 40
datasets. The results show that the proposed techniques achieved interesting
results, with KNORA-BI outperforming state-of-art techniques.
|
Distributed Simulation and Distributed Inference | Independent samples from an unknown probability distribution $\bf p$ on a
domain of size $k$ are distributed across $n$ players, with each player holding
one sample. Each player can communicate $\ell$ bits to a central referee in a
simultaneous message passing model of communication to help the referee infer a
property of the unknown $\bf p$. What is the least number of players for
inference required in the communication-starved setting of $\ell<\log k$? We
begin by exploring a general "simulate-and-infer" strategy for such inference
problems where the center simulates the desired number of samples from the
unknown distribution and applies standard inference algorithms for the
collocated setting. Our first result shows that for $\ell<\log k$ perfect
simulation of even a single sample is not possible. Nonetheless, we present a
Las Vegas algorithm that simulates a single sample from the unknown
distribution using $O(k/2^\ell)$ samples in expectation. As an immediate
corollary, we get that simulate-and-infer attains the optimal sample complexity
of $\Theta(k^2/2^\ell\epsilon^2)$ for learning the unknown distribution to
total variation distance $\epsilon$. For the prototypical testing problem of
identity testing, simulate-and-infer works with $O(k^{3/2}/2^\ell\epsilon^2)$
samples, a requirement that seems to be inherent for all communication
protocols not using any additional resources. Interestingly, we can break this
barrier using public coins. Specifically, we exhibit a public-coin
communication protocol that performs identity testing using
$O(k/\sqrt{2^\ell}\epsilon^2)$ samples. Furthermore, we show that this is
optimal up to constant factors. Our theoretically sample-optimal protocol is
easy to implement in practice. Our proof of lower bound entails showing a
contraction in $\chi^2$ distance of product distributions due to communication
constraints and may be of independent interest.
|
GNAS: A Greedy Neural Architecture Search Method for Multi-Attribute
Learning | A key problem in deep multi-attribute learning is to effectively discover the
inter-attribute correlation structures. Typically, the conventional deep
multi-attribute learning approaches follow the pipeline of manually designing
the network architectures based on task-specific expertise prior knowledge and
careful network tunings, leading to the inflexibility for various complicated
scenarios in practice. Motivated by addressing this problem, we propose an
efficient greedy neural architecture search approach (GNAS) to automatically
discover the optimal tree-like deep architecture for multi-attribute learning.
In a greedy manner, GNAS divides the optimization of global architecture into
the optimizations of individual connections step by step. By iteratively
updating the local architectures, the global tree-like architecture gets
converged where the bottom layers are shared across relevant attributes and the
branches in top layers more encode attribute-specific features. Experiments on
three benchmark multi-attribute datasets show the effectiveness and compactness
of neural architectures derived by GNAS, and also demonstrate the efficiency of
GNAS in searching neural architectures.
|
Forward-Backward Stochastic Neural Networks: Deep Learning of
High-dimensional Partial Differential Equations | Classical numerical methods for solving partial differential equations suffer
from the curse dimensionality mainly due to their reliance on meticulously
generated spatio-temporal grids. Inspired by modern deep learning based
techniques for solving forward and inverse problems associated with partial
differential equations, we circumvent the tyranny of numerical discretization
by devising an algorithm that is scalable to high-dimensions. In particular, we
approximate the unknown solution by a deep neural network which essentially
enables us to benefit from the merits of automatic differentiation. To train
the aforementioned neural network we leverage the well-known connection between
high-dimensional partial differential equations and forward-backward stochastic
differential equations. In fact, independent realizations of a standard
Brownian motion will act as training data. We test the effectiveness of our
approach for a couple of benchmark problems spanning a number of scientific
domains including Black-Scholes-Barenblatt and Hamilton-Jacobi-Bellman
equations, both in 100-dimensions.
|
Semantic Adversarial Deep Learning | Fueled by massive amounts of data, models produced by machine-learning (ML)
algorithms, especially deep neural networks, are being used in diverse domains
where trustworthiness is a concern, including automotive systems, finance,
health care, natural language processing, and malware detection. Of particular
concern is the use of ML algorithms in cyber-physical systems (CPS), such as
self-driving cars and aviation, where an adversary can cause serious
consequences. However, existing approaches to generating adversarial examples
and devising robust ML algorithms mostly ignore the semantics and context of
the overall system containing the ML component. For example, in an autonomous
vehicle using deep learning for perception, not every adversarial example for
the neural network might lead to a harmful consequence. Moreover, one may want
to prioritize the search for adversarial examples towards those that
significantly modify the desired semantics of the overall system. Along the
same lines, existing algorithms for constructing robust ML algorithms ignore
the specification of the overall system. In this paper, we argue that the
semantics and specification of the overall system has a crucial role to play in
this line of research. We present preliminary research results that support
this claim.
|
Exploring Partially Observed Networks with Nonparametric Bandits | Real-world networks such as social and communication networks are too large
to be observed entirely. Such networks are often partially observed such that
network size, network topology, and nodes of the original network are unknown.
In this paper we formalize the Adaptive Graph Exploring problem. We assume that
we are given an incomplete snapshot of a large network and additional nodes can
be discovered by querying nodes in the currently observed network. The goal of
this problem is to maximize the number of observed nodes within a given query
budget. Querying which set of nodes maximizes the size of the observed network?
We formulate this problem as an exploration-exploitation problem and propose a
novel nonparametric multi-arm bandit (MAB) algorithm for identifying which
nodes to be queried. Our contributions include: (1) $i$KNN-UCB, a novel
nonparametric MAB algorithm, applies $k$-nearest neighbor UCB to the setting
when the arms are presented in a vector space, (2) provide theoretical
guarantee that $i$KNN-UCB algorithm has sublinear regret, and (3) applying
$i$KNN-UCB algorithm on synthetic networks and real-world networks from
different domains, we show that our method discovers up to 40% more nodes
compared to existing baselines.
|
Robustness via Deep Low-Rank Representations | We investigate the effect of the dimensionality of the representations
learned in Deep Neural Networks (DNNs) on their robustness to input
perturbations, both adversarial and random. To achieve low dimensionality of
learned representations, we propose an easy-to-use, end-to-end trainable,
low-rank regularizer (LR) that can be applied to any intermediate layer
representation of a DNN. This regularizer forces the feature representations to
(mostly) lie in a low-dimensional linear subspace. We perform a wide range of
experiments that demonstrate that the LR indeed induces low rank on the
representations, while providing modest improvements to accuracy as an added
benefit. Furthermore, the learned features make the trained model significantly
more robust to input perturbations such as Gaussian and adversarial noise (even
without adversarial training). Lastly, the low-dimensionality means that the
learned features are highly compressible; thus discriminative features of the
data can be stored using very little memory. Our experiments indicate that
models trained using the LR learn robust classifiers by discovering subspaces
that avoid non-robust features. Algorithmically, the LR is scalable, generic,
and straightforward to implement into existing deep learning frameworks.
|
Detecting Regions of Maximal Divergence for Spatio-Temporal Anomaly
Detection | Automatic detection of anomalies in space- and time-varying measurements is
an important tool in several fields, e.g., fraud detection, climate analysis,
or healthcare monitoring. We present an algorithm for detecting anomalous
regions in multivariate spatio-temporal time-series, which allows for spotting
the interesting parts in large amounts of data, including video and text data.
In opposition to existing techniques for detecting isolated anomalous data
points, we propose the "Maximally Divergent Intervals" (MDI) framework for
unsupervised detection of coherent spatial regions and time intervals
characterized by a high Kullback-Leibler divergence compared with all other
data given. In this regard, we define an unbiased Kullback-Leibler divergence
that allows for ranking regions of different size and show how to enable the
algorithm to run on large-scale data sets in reasonable time using an interval
proposal technique. Experiments on both synthetic and real data from various
domains, such as climate analysis, video surveillance, and text forensics,
demonstrate that our method is widely applicable and a valuable tool for
finding interesting events in different types of data.
|
Unsupervised Prostate Cancer Detection on H&E using Convolutional
Adversarial Autoencoders | We propose an unsupervised method using self-clustering convolutional
adversarial autoencoders to classify prostate tissue as tumor or non-tumor
without any labeled training data. The clustering method is integrated into the
training of the autoencoder and requires only little post-processing. Our
network trains on hematoxylin and eosin (H&E) input patches and we tested two
different reconstruction targets, H&E and immunohistochemistry (IHC). We show
that antibody-driven feature learning using IHC helps the network to learn
relevant features for the clustering task. Our network achieves a F1 score of
0.62 using only a small set of validation labels to assign classes to clusters.
|
Dictionary learning -- from local towards global and adaptive | This paper studies the convergence behaviour of dictionary learning via the
Iterative Thresholding and K-residual Means (ITKrM) algorithm. On one hand it
is proved that ITKrM is a contraction under much more relaxed conditions than
previously necessary. On the other hand it is shown that there seem to exist
stable fixed points that do not correspond to the generating dictionary, which
can be characterised as very coherent. Based on an analysis of the residuals
using these bad dictionaries, replacing coherent atoms with carefully designed
replacement candidates is proposed. In experiments on synthetic data this
outperforms random or no replacement and always leads to full dictionary
recovery. Finally the question how to learn dictionaries without knowledge of
the correct dictionary size and sparsity level is addressed. Decoupling the
replacement strategy of coherent or unused atoms into pruning and adding, and
slowly carefully increasing the sparsity level, leads to an adaptive version of
ITKrM. In several experiments this adaptive dictionary learning algorithm is
shown to recover a generating dictionary from randomly initialised dictionaries
of various sizes on synthetic data and to learn meaningful dictionaries on
image data.
|
varrank: an R package for variable ranking based on mutual information
with applications to observed systemic datasets | This article describes the R package varrank. It has a flexible
implementation of heuristic approaches which perform variable ranking based on
mutual information. The package is particularly suitable for exploring
multivariate datasets requiring a holistic analysis. The core functionality is
a general implementation of the minimum redundancy maximum relevance (mRMRe)
model. This approach is based on information theory metrics. It is compatible
with discrete and continuous data which are discretised using a large choice of
possible rules. The two main problems that can be addressed by this package are
the selection of the most representative variables for modeling a collection of
variables of interest, i.e., dimension reduction, and variable ranking with
respect to a set of variables of interest.
|
Human Activity Recognition using Recurrent Neural Networks | Human activity recognition using smart home sensors is one of the bases of
ubiquitous computing in smart environments and a topic undergoing intense
research in the field of ambient assisted living. The increasingly large amount
of data sets calls for machine learning methods. In this paper, we introduce a
deep learning model that learns to classify human activities without using any
prior knowledge. For this purpose, a Long Short Term Memory (LSTM) Recurrent
Neural Network was applied to three real world smart home datasets. The results
of these experiments show that the proposed approach outperforms the existing
ones in terms of accuracy and performance.
|
Scalable attribute-aware network embedding with locality | Adding attributes for nodes to network embedding helps to improve the ability
of the learned joint representation to depict features from topology and
attributes simultaneously. Recent research on the joint embedding has exhibited
a promising performance on a variety of tasks by jointly embedding the two
spaces. However, due to the indispensable requirement of globality based
information, present approaches contain a flaw of in-scalability. Here we
propose \emph{SANE}, a scalable attribute-aware network embedding algorithm
with locality, to learn the joint representation from topology and attributes.
By enforcing the alignment of a local linear relationship between each node and
its K-nearest neighbors in topology and attribute space, the joint embedding
representations are more informative comparing with a single representation
from topology or attributes alone. And we argue that the locality in
\emph{SANE} is the key to learning the joint representation at scale. By using
several real-world networks from diverse domains, We demonstrate the efficacy
of \emph{SANE} in performance and scalability aspect. Overall, for performance
on label classification, SANE successfully reaches up to the highest F1-score
on most datasets, and even closer to the baseline method that needs label
information as extra inputs, compared with other state-of-the-art joint
representation algorithms. What's more, \emph{SANE} has an up to 71.4\%
performance gain compared with the single topology-based algorithm. For
scalability, we have demonstrated the linearly time complexity of \emph{SANE}.
In addition, we intuitively observe that when the network size scales to
100,000 nodes, the "learning joint embedding" step of \emph{SANE} only takes
$\approx10$ seconds.
|
Instance Selection Improves Geometric Mean Accuracy: A Study on
Imbalanced Data Classification | A natural way of handling imbalanced data is to attempt to equalise the class
frequencies and train the classifier of choice on balanced data. For two-class
imbalanced problems, the classification success is typically measured by the
geometric mean (GM) of the true positive and true negative rates. Here we prove
that GM can be improved upon by instance selection, and give the theoretical
conditions for such an improvement. We demonstrate that GM is non-monotonic
with respect to the number of retained instances, which discourages systematic
instance selection. We also show that balancing the distribution frequencies is
inferior to a direct maximisation of GM. To verify our theoretical findings, we
carried out an experimental study of 12 instance selection methods for
imbalanced data, using 66 standard benchmark data sets. The results reveal
possible room for new instance selection methods for imbalanced data.
|
Large-scale Nonlinear Variable Selection via Kernel Random Features | We propose a new method for input variable selection in nonlinear regression.
The method is embedded into a kernel regression machine that can model general
nonlinear functions, not being a priori limited to additive models. This is the
first kernel-based variable selection method applicable to large datasets. It
sidesteps the typical poor scaling properties of kernel methods by mapping the
inputs into a relatively low-dimensional space of random features. The
algorithm discovers the variables relevant for the regression task together
with learning the prediction model through learning the appropriate nonlinear
random feature maps. We demonstrate the outstanding performance of our method
on a set of large-scale synthetic and real datasets.
|
Lipschitz Continuity in Model-based Reinforcement Learning | We examine the impact of learning Lipschitz continuous models in the context
of model-based reinforcement learning. We provide a novel bound on multi-step
prediction error of Lipschitz models where we quantify the error using the
Wasserstein metric. We go on to prove an error bound for the value-function
estimate arising from Lipschitz models and show that the estimated value
function is itself Lipschitz. We conclude with empirical results that show the
benefits of controlling the Lipschitz constant of neural-network models.
|
NAIS-Net: Stable Deep Networks from Non-Autonomous Differential
Equations | This paper introduces Non-Autonomous Input-Output Stable Network(NAIS-Net), a
very deep architecture where each stacked processing block is derived from a
time-invariant non-autonomous dynamical system. Non-autonomy is implemented by
skip connections from the block input to each of the unrolled processing stages
and allows stability to be enforced so that blocks can be unrolled adaptively
to a pattern-dependent processing depth. NAIS-Net induces non-trivial,
Lipschitz input-output maps, even for an infinite unroll length. We prove that
the network is globally asymptotically stable so that for every initial
condition there is exactly one input-dependent equilibrium assuming $tanh$
units, and incrementally stable for ReL units. An efficient implementation that
enforces the stability under derived conditions for both fully-connected and
convolutional layers is also presented. Experimental results show how NAIS-Net
exhibits stability in practice, yielding a significant reduction in
generalization gap compared to ResNets.
|
Multi-view Hybrid Embedding: A Divide-and-Conquer Approach | We present a novel cross-view classification algorithm where the gallery and
probe data come from different views. A popular approach to tackle this problem
is the multi-view subspace learning (MvSL) that aims to learn a latent subspace
shared by multi-view data. Despite promising results obtained on some
applications, the performance of existing methods deteriorates dramatically
when the multi-view data is sampled from nonlinear manifolds or suffers from
heavy outliers. To circumvent this drawback, motivated by the
Divide-and-Conquer strategy, we propose Multi-view Hybrid Embedding (MvHE), a
unique method of dividing the problem of cross-view classification into three
subproblems and building one model for each subproblem. Specifically, the first
model is designed to remove view discrepancy, whereas the second and third
models attempt to discover the intrinsic nonlinear structure and to increase
discriminability in intra-view and inter-view samples respectively. The kernel
extension is conducted to further boost the representation power of MvHE.
Extensive experiments are conducted on four benchmark datasets. Our methods
demonstrate overwhelming advantages against the state-of-the-art MvSL based
cross-view classification approaches in terms of classification accuracy and
robustness.
|
A sequential sampling strategy for extreme event statistics in nonlinear
dynamical systems | We develop a method for the evaluation of extreme event statistics associated
with nonlinear dynamical systems, using a small number of samples. From an
initial dataset of design points, we formulate a sequential strategy that
provides the 'next-best' data point (set of parameters) that when evaluated
results in improved estimates of the probability density function (pdf) for a
scalar quantity of interest. The approach utilizes Gaussian process regression
to perform Bayesian inference on the parameter-to-observation map describing
the quantity of interest. We then approximate the desired pdf along with
uncertainty bounds utilizing the posterior distribution of the inferred map.
The 'next-best' design point is sequentially determined through an optimization
procedure that selects the point in parameter space that maximally reduces
uncertainty between the estimated bounds of the pdf prediction. Since the
optimization process utilizes only information from the inferred map it has
minimal computational cost. Moreover, the special form of the metric emphasizes
the tails of the pdf. The method is practical for systems where the
dimensionality of the parameter space is of moderate size, i.e. order O(10). We
apply the method to estimate the extreme event statistics for a very
high-dimensional system with millions of degrees of freedom: an offshore
platform subjected to three-dimensional irregular waves. It is demonstrated
that the developed approach can accurately determine the extreme event
statistics using limited number of samples.
|
Deep Transfer Network with Joint Distribution Adaptation: A New
Intelligent Fault Diagnosis Framework for Industry Application | In recent years, an increasing popularity of deep learning model for
intelligent condition monitoring and diagnosis as well as prognostics used for
mechanical systems and structures has been observed. In the previous studies,
however, a major assumption accepted by default, is that the training and
testing data are taking from same feature distribution. Unfortunately, this
assumption is mostly invalid in real application, resulting in a certain lack
of applicability for the traditional diagnosis approaches. Inspired by the idea
of transfer learning that leverages the knowledge learnt from rich labeled data
in source domain to facilitate diagnosing a new but similar target task, a new
intelligent fault diagnosis framework, i.e., deep transfer network (DTN), which
generalizes deep learning model to domain adaptation scenario, is proposed in
this paper. By extending the marginal distribution adaptation (MDA) to joint
distribution adaptation (JDA), the proposed framework can exploit the
discrimination structures associated with the labeled data in source domain to
adapt the conditional distribution of unlabeled target data, and thus guarantee
a more accurate distribution matching. Extensive empirical evaluations on three
fault datasets validate the applicability and practicability of DTN, while
achieving many state-of-the-art transfer results in terms of diverse operating
conditions, fault severities and fault types.
|
Socially Guided Intrinsic Motivation for Robot Learning of Motor Skills | This paper presents a technical approach to robot learning of motor skills
which combines active intrinsically motivated learning with imitation learning.
Our architecture, called SGIM-D, allows efficient learning of high-dimensional
continuous sensorimotor inverse models in robots, and in particular learns
distributions of parameterised motor policies that solve a corresponding
distribution of parameterised goals/tasks. This is made possible by the
technical integration of imitation learning techniques within an algorithm for
learning inverse models that relies on active goal babbling. After reviewing
social learning and intrinsic motivation approaches to action learning, we
describe the general framework of our algorithm, before detailing its
architecture. In an experiment where a robot arm has to learn to use a flexible
fishing line , we illustrate that SGIM-D efficiently combines the advantages of
social learning and intrinsic motivation and benefits from human demonstration
properties to learn how to produce varied outcomes in the environment, while
developing more precise control policies in large spaces.
|
Deep Dynamic Boosted Forest | Random forest is widely exploited as an ensemble learning method. In many
practical applications, however, there is still a significant challenge to
learn from imbalanced data. To alleviate this limitation, we propose a deep
dynamic boosted forest (DDBF), a novel ensemble algorithm that incorporates the
notion of hard example mining into random forest. Specically, we propose to
measure the quality of each leaf node of every decision tree in the random
forest to determine hard examples. By iteratively training and then removing
easy examples from training data, we evolve the random forest to focus on hard
examples dynamically so as to balance the proportion of samples and learn
decision boundaries better. Data can be cascaded through these random forests
learned in each iteration in sequence to generate more accurate predictions.
Our DDBF outperforms random forest on 5 UCI datasets, MNIST and SATIMAGE, and
achieved state-of-the-art results compared to other deep models. Moreover, we
show that DDBF is also a new way of sampling and can be very useful and
efficient when learning from imbalanced data.
|
Deep Triplet Ranking Networks for One-Shot Recognition | Despite the breakthroughs achieved by deep learning models in conventional
supervised learning scenarios, their dependence on sufficient labeled training
data in each class prevents effective applications of these deep models in
situations where labeled training instances for a subset of novel classes are
very sparse -- in the extreme case only one instance is available for each
class. To tackle this natural and important challenge, one-shot learning, which
aims to exploit a set of well labeled base classes to build classifiers for the
new target classes that have only one observed instance per class, has recently
received increasing attention from the research community. In this paper we
propose a novel end-to-end deep triplet ranking network to perform one-shot
learning. The proposed approach learns class universal image embeddings on the
well labeled base classes under a triplet ranking loss, such that the instances
from new classes can be categorized based on their similarity with the one-shot
instances in the learned embedding space. Moreover, our approach can naturally
incorporate the available one-shot instances from the new classes into the
embedding learning process to improve the triplet ranking model. We conduct
experiments on two popular datasets for one-shot learning. The results show the
proposed approach achieves better performance than the state-of-the- art
comparison methods.
|
Generating Music using an LSTM Network | A model of music needs to have the ability to recall past details and have a
clear, coherent understanding of musical structure. Detailed in the paper is a
neural network architecture that predicts and generates polyphonic music
aligned with musical rules. The probabilistic model presented is a Bi-axial
LSTM trained with a kernel reminiscent of a convolutional kernel. When analyzed
quantitatively and qualitatively, this approach performs well in composing
polyphonic music. Link to the code is provided.
|
Nonparametric Stochastic Compositional Gradient Descent for Q-Learning
in Continuous Markov Decision Problems | We consider Markov Decision Problems defined over continuous state and action
spaces, where an autonomous agent seeks to learn a map from its states to
actions so as to maximize its long-term discounted accumulation of rewards. We
address this problem by considering Bellman's optimality equation defined over
action-value functions, which we reformulate into a nested non-convex
stochastic optimization problem defined over a Reproducing Kernel Hilbert Space
(RKHS). We develop a functional generalization of stochastic quasi-gradient
method to solve it, which, owing to the structure of the RKHS, admits a
parameterization in terms of scalar weights and past state-action pairs which
grows proportionately with the algorithm iteration index. To ameliorate this
complexity explosion, we apply Kernel Orthogonal Matching Pursuit to the
sequence of kernel weights and dictionaries, which yields a controllable error
in the descent direction of the underlying optimization method. We prove that
the resulting algorithm, called KQ-Learning, converges with probability 1 to a
stationary point of this problem, yielding a fixed point of the Bellman
optimality operator under the hypothesis that it belongs to the RKHS. Under
constant learning rates, we further obtain convergence to a small Bellman error
that depends on the chosen learning rates. Numerical evaluation on the
Continuous Mountain Car and Inverted Pendulum tasks yields convergent
parsimonious learned action-value functions, policies that are competitive with
the state of the art, and exhibit reliable, reproducible learning behavior.
|
Effects of sampling skewness of the importance-weighted risk estimator
on model selection | Importance-weighting is a popular and well-researched technique for dealing
with sample selection bias and covariate shift. It has desirable
characteristics such as unbiasedness, consistency and low computational
complexity. However, weighting can have a detrimental effect on an estimator as
well. In this work, we empirically show that the sampling distribution of an
importance-weighted estimator can be skewed. For sample selection bias
settings, and for small sample sizes, the importance-weighted risk estimator
produces overestimates for datasets in the body of the sampling distribution,
i.e. the majority of cases, and large underestimates for data sets in the tail
of the sampling distribution. These over- and underestimates of the risk lead
to suboptimal regularization parameters when used for importance-weighted
validation.
|
Randomized ICA and LDA Dimensionality Reduction Methods for
Hyperspectral Image Classification | Dimensionality reduction is an important step in processing the hyperspectral
images (HSI) to overcome the curse of dimensionality problem. Linear
dimensionality reduction methods such as Independent component analysis (ICA)
and Linear discriminant analysis (LDA) are commonly employed to reduce the
dimensionality of HSI. These methods fail to capture non-linear dependency in
the HSI data, as data lies in the nonlinear manifold. To handle this, nonlinear
transformation techniques based on kernel methods were introduced for
dimensionality reduction of HSI. However, the kernel methods involve cubic
computational complexity while computing the kernel matrix, and thus its
potential cannot be explored when the number of pixels (samples) are large. In
literature a fewer number of pixels are randomly selected to partial to
overcome this issue, however this sub-optimal strategy might neglect important
information in the HSI. In this paper, we propose randomized solutions to the
ICA and LDA dimensionality reduction methods using Random Fourier features, and
we label them as RFFICA and RFFLDA. Our proposed method overcomes the
scalability issue and to handle the non-linearities present in the data more
efficiently. Experiments conducted with two real-world hyperspectral datasets
demonstrates that our proposed randomized methods outperform the conventional
kernel ICA and kernel LDA in terms overall, per-class accuracies and
computational time.
|
Sampling-free Uncertainty Estimation in Gated Recurrent Units with
Exponential Families | There has recently been a concerted effort to derive mechanisms in vision and
machine learning systems to offer uncertainty estimates of the predictions they
make. Clearly, there are enormous benefits to a system that is not only
accurate but also has a sense for when it is not sure. Existing proposals
center around Bayesian interpretations of modern deep architectures -- these
are effective but can often be computationally demanding. We show how classical
ideas in the literature on exponential families on probabilistic networks
provide an excellent starting point to derive uncertainty estimates in Gated
Recurrent Units (GRU). Our proposal directly quantifies uncertainty
deterministically, without the need for costly sampling-based estimation. We
demonstrate how our model can be used to quantitatively and qualitatively
measure uncertainty in unsupervised image sequence prediction. To our
knowledge, this is the first result describing sampling-free uncertainty
estimation for powerful sequential models such as GRUs.
|
Unsupervised Representation Adversarial Learning Network: from
Reconstruction to Generation | A good representation for arbitrarily complicated data should have the
capability of semantic generation, clustering and reconstruction. Previous
research has already achieved impressive performance on either one. This paper
aims at learning a disentangled representation effective for all of them in an
unsupervised way. To achieve all the three tasks together, we learn the forward
and inverse mapping between data and representation on the basis of a symmetric
adversarial process. In theory, we minimize the upper bound of the two
conditional entropy loss between the latent variables and the observations
together to achieve the cycle consistency. The newly proposed RepGAN is tested
on MNIST, fashionMNIST, CelebA, and SVHN datasets to perform unsupervised
classification, generation and reconstruction tasks. The result demonstrates
that RepGAN is able to learn a useful and competitive representation. To the
author's knowledge, our work is the first one to achieve both a high
unsupervised classification accuracy and low reconstruction error on MNIST.
Codes are available at https://github.com/yzhouas/RepGAN-tensorflow.
|
GritNet: Student Performance Prediction with Deep Learning | Student performance prediction - where a machine forecasts the future
performance of students as they interact with online coursework - is a
challenging problem. Reliable early-stage predictions of a student's future
performance could be critical to facilitate timely educational interventions
during a course. However, very few prior studies have explored this problem
from a deep learning perspective. In this paper, we recast the student
performance prediction problem as a sequential event prediction problem and
propose a new deep learning based algorithm, termed GritNet, which builds upon
the bidirectional long short term memory (BLSTM). Our results, from real
Udacity students' graduation predictions, show that the GritNet not only
consistently outperforms the standard logistic-regression based method, but
that improvements are substantially pronounced in the first few weeks when
accurate predictions are most challenging.
|
An Ensemble Generation Method Based on Instance Hardness | In Machine Learning, ensemble methods have been receiving a great deal of
attention. Techniques such as Bagging and Boosting have been successfully
applied to a variety of problems. Nevertheless, such techniques are still
susceptible to the effects of noise and outliers in the training data. We
propose a new method for the generation of pools of classifiers based on
Bagging, in which the probability of an instance being selected during the
resampling process is inversely proportional to its instance hardness, which
can be understood as the likelihood of an instance being misclassified,
regardless of the choice of classifier. The goal of the proposed method is to
remove noisy data without sacrificing the hard instances which are likely to be
found on class boundaries. We evaluate the performance of the method in
nineteen public data sets, and compare it to the performance of the Bagging and
Random Subspace algorithms. Our experiments show that in high noise scenarios
the accuracy of our method is significantly better than that of Bagging.
|
Two Use Cases of Machine Learning for SDN-Enabled IP/Optical Networks:
Traffic Matrix Prediction and Optical Path Performance Prediction | We describe two applications of machine learning in the context of IP/Optical
networks. The first one allows agile management of resources at a core
IP/Optical network by using machine learning for short-term and long-term
prediction of traffic flows and joint global optimization of IP and optical
layers using colorless/directionless (CD) flexible ROADMs. Multilayer
coordination allows for significant cost savings, flexible new services to meet
dynamic capacity needs, and improved robustness by being able to proactively
adapt to new traffic patterns and network conditions. The second application is
important as we migrate our metro networks to Open ROADM networks, to allow
physical routing without the need for detailed knowledge of optical parameters.
We discuss a proof-of-concept study, where detailed performance data for
wavelengths on a current flexible ROADM network is used for machine learning to
predict the optical performance of each wavelength. Both applications can be
efficiently implemented by using a SDN (Software Defined Network) controller.
|
Streaming Active Learning Strategies for Real-Life Credit Card Fraud
Detection: Assessment and Visualization | Credit card fraud detection is a very challenging problem because of the
specific nature of transaction data and the labeling process. The transaction
data is peculiar because they are obtained in a streaming fashion, they are
strongly imbalanced and prone to non-stationarity. The labeling is the outcome
of an active learning process, as every day human investigators contact only a
small number of cardholders (associated to the riskiest transactions) and
obtain the class (fraud or genuine) of the related transactions. An adequate
selection of the set of cardholders is therefore crucial for an efficient fraud
detection process. In this paper, we present a number of active learning
strategies and we investigate their fraud detection accuracies. We compare
different criteria (supervised, semi-supervised and unsupervised) to query
unlabeled transactions. Finally, we highlight the existence of an
exploitation/exploration trade-off for active learning in the context of fraud
detection, which has so far been overlooked in the literature.
|
MobileFaceNets: Efficient CNNs for Accurate Real-Time Face Verification
on Mobile Devices | We present a class of extremely efficient CNN models, MobileFaceNets, which
use less than 1 million parameters and are specifically tailored for
high-accuracy real-time face verification on mobile and embedded devices. We
first make a simple analysis on the weakness of common mobile networks for face
verification. The weakness has been well overcome by our specifically designed
MobileFaceNets. Under the same experimental conditions, our MobileFaceNets
achieve significantly superior accuracy as well as more than 2 times actual
speedup over MobileNetV2. After trained by ArcFace loss on the refined
MS-Celeb-1M, our single MobileFaceNet of 4.0MB size achieves 99.55% accuracy on
LFW and 92.59% TAR@FAR1e-6 on MegaFace, which is even comparable to
state-of-the-art big CNN models of hundreds MB size. The fastest one of
MobileFaceNets has an actual inference time of 18 milliseconds on a mobile
phone. For face verification, MobileFaceNets achieve significantly improved
efficiency over previous state-of-the-art mobile CNNs.
|
Robust And Scalable Learning Of Complex Dataset Topologies Via Elpigraph | Large datasets represented by multidimensional data point clouds often
possess non-trivial distributions with branching trajectories and excluded
regions, with the recent single-cell transcriptomic studies of developing
embryo being notable examples. Reducing the complexity and producing compact
and interpretable representations of such data remains a challenging task. Most
of the existing computational methods are based on exploring the local data
point neighbourhood relations, a step that can perform poorly in the case of
multidimensional and noisy data. Here we present ElPiGraph, a scalable and
robust method for approximation of datasets with complex structures which does
not require computing the complete data distance matrix or the data point
neighbourhood graph. This method is able to withstand high levels of noise and
is capable of approximating complex topologies via principal graph ensembles
that can be combined into a consensus principal graph. ElPiGraph deals
efficiently with large and complex datasets in various fields from biology,
where it can be used to infer gene dynamics from single-cell RNA-Seq, to
astronomy, where it can be used to explore complex structures in the
distribution of galaxies.
|
Revisiting Small Batch Training for Deep Neural Networks | Modern deep neural network training is typically based on mini-batch
stochastic gradient optimization. While the use of large mini-batches increases
the available computational parallelism, small batch training has been shown to
provide improved generalization performance and allows a significantly smaller
memory footprint, which might also be exploited to improve machine throughput.
In this paper, we review common assumptions on learning rate scaling and
training duration, as a basis for an experimental comparison of test
performance for different mini-batch sizes. We adopt a learning rate that
corresponds to a constant average weight update per gradient calculation (i.e.,
per unit cost of computation), and point out that this results in a variance of
the weight updates that increases linearly with the mini-batch size $m$.
The collected experimental results for the CIFAR-10, CIFAR-100 and ImageNet
datasets show that increasing the mini-batch size progressively reduces the
range of learning rates that provide stable convergence and acceptable test
performance. On the other hand, small mini-batch sizes provide more up-to-date
gradient calculations, which yields more stable and reliable training. The best
performance has been consistently obtained for mini-batch sizes between $m = 2$
and $m = 32$, which contrasts with recent work advocating the use of mini-batch
sizes in the thousands.
|
A Simple Quantum Neural Net with a Periodic Activation Function | In this paper, we propose a simple neural net that requires only $O(nlog_2k)$
number of qubits and $O(nk)$ quantum gates: Here, $n$ is the number of input
parameters, and $k$ is the number of weights applied to these parameters in the
proposed neural net. We describe the network in terms of a quantum circuit, and
then draw its equivalent classical neural net which involves $O(k^n)$ nodes in
the hidden layer. Then, we show that the network uses a periodic activation
function of cosine values of the linear combinations of the inputs and weights.
The backpropagation is described through the gradient descent, and then iris
and breast cancer datasets are used for the simulations. The numerical results
indicate the network can be used in machine learning problems and it may
provide exponential speedup over the same structured classical neural net.
|
One-Shot Learning using Mixture of Variational Autoencoders: a
Generalization Learning approach | Deep learning, even if it is very successful nowadays, traditionally needs
very large amounts of labeled data to perform excellent on the classification
task. In an attempt to solve this problem, the one-shot learning paradigm,
which makes use of just one labeled sample per class and prior knowledge,
becomes increasingly important. In this paper, we propose a new one-shot
learning method, dubbed MoVAE (Mixture of Variational AutoEncoders), to perform
classification. Complementary to prior studies, MoVAE represents a shift of
paradigm in comparison with the usual one-shot learning methods, as it does not
use any prior knowledge. Instead, it starts from zero knowledge and one labeled
sample per class. Afterward, by using unlabeled data and the generalization
learning concept (in a way, more as humans do), it is capable to gradually
improve by itself its performance. Even more, if there are no unlabeled data
available MoVAE can still perform well in one-shot learning classification. We
demonstrate empirically the efficiency of our proposed approach on three
datasets, i.e. the handwritten digits (MNIST), fashion products
(Fashion-MNIST), and handwritten characters (Omniglot), showing that MoVAE
outperforms state-of-the-art one-shot learning algorithms.
|
Modelling customer online behaviours with neural networks: applications
to conversion prediction and advertising retargeting | In this paper, we apply neural networks into digital marketing world for the
purpose of better targeting the potential customers. To do so, we model the
customer online behaviours using dedicated neural network architectures.
Starting from user searched keywords in a search engine to the landing page and
different following pages, until the user left the site, we model the whole
visited journey with a Recurrent Neural Network (RNN), together with
Convolution Neural Networks (CNN) that can take into account of the semantic
meaning of user searched keywords and different visited page names. With such
model, we use Monte Carlo simulation to estimate the conversion rates of each
potential customer in the future visiting. We believe our concept and the
preliminary promising results in this paper enable the use of largely available
customer online behaviours data for advanced digital marketing analysis.
|
Unsupervised learning of the brain connectivity dynamic using residual
D-net | In this paper, we propose a novel unsupervised learning method to learn the
brain dynamics using a deep learning architecture named residual D-net. As it
is often the case in medical research, in contrast to typical deep learning
tasks, the size of the resting-state functional Magnetic Resonance Image
(rs-fMRI) datasets for training is limited. Thus, the available data should be
very efficiently used to learn the complex patterns underneath the brain
connectivity dynamics. To address this issue, we use residual connections to
alleviate the training complexity through recurrent multi-scale representation.
We conduct two classification tasks to differentiate early and late stage Mild
Cognitive Impairment (MCI) from Normal healthy Control (NC) subjects. The
experiments verify that our proposed residual D-net indeed learns the brain
connectivity dynamics, leading to significantly higher classification accuracy
compared to previously published techniques.
|
ADef: an Iterative Algorithm to Construct Adversarial Deformations | While deep neural networks have proven to be a powerful tool for many
recognition and classification tasks, their stability properties are still not
well understood. In the past, image classifiers have been shown to be
vulnerable to so-called adversarial attacks, which are created by additively
perturbing the correctly classified image. In this paper, we propose the ADef
algorithm to construct a different kind of adversarial attack created by
iteratively applying small deformations to the image, found through a gradient
descent step. We demonstrate our results on MNIST with convolutional neural
networks and on ImageNet with Inception-v3 and ResNet-101.
|
Loss in Translation: Learning Bilingual Word Mapping with a Retrieval
Criterion | Continuous word representations learned separately on distinct languages can
be aligned so that their words become comparable in a common space. Existing
works typically solve a least-square regression problem to learn a rotation
aligning a small bilingual lexicon, and use a retrieval criterion for
inference. In this paper, we propose an unified formulation that directly
optimizes a retrieval criterion in an end-to-end fashion. Our experiments on
standard benchmarks show that our approach outperforms the state of the art on
word translation, with the biggest improvements observed for distant language
pairs such as English-Chinese.
|
Learning More Robust Features with Adversarial Training | In recent years, it has been found that neural networks can be easily fooled
by adversarial examples, which is a potential safety hazard in some
safety-critical applications. Many researchers have proposed various method to
make neural networks more robust to white-box adversarial attacks, but an
effective method have not been found so far. In this short paper, we focus on
the robustness of the features learned by neural networks. We show that the
features learned by neural networks are not robust, and find that the
robustness of the learned features is closely related to the resistance against
adversarial examples of neural networks. We also find that adversarial training
against fast gradients sign method (FGSM) does not make the leaned features
very robust, even if it can make the trained networks very resistant to FGSM
attack. Then we propose a method, which can be seen as an extension of
adversarial training, to train neural networks to learn more robust features.
We perform experiments on MNIST and CIFAR-10 to evaluate our method, and the
experiment results show that this method greatly improves the robustness of the
learned features and the resistance to adversarial attacks.
|
Mapping Images to Psychological Similarity Spaces Using Neural Networks | The cognitive framework of conceptual spaces bridges the gap between symbolic
and subsymbolic AI by proposing an intermediate conceptual layer where
knowledge is represented geometrically. There are two main approaches for
obtaining the dimensions of this conceptual similarity space: using similarity
ratings from psychological experiments and using machine learning techniques.
In this paper, we propose a combination of both approaches by using
psychologically derived similarity ratings to constrain the machine learning
process. This way, a mapping from stimuli to conceptual spaces can be learned
that is both supported by psychological data and allows generalization to
unseen stimuli. The results of a first feasibility study support our proposed
approach.
|
A Self-paced Regularization Framework for Partial-Label Learning | Partial label learning (PLL) aims to solve the problem where each training
instance is associated with a set of candidate labels, one of which is the
correct label. Most PLL algorithms try to disambiguate the candidate label set,
by either simply treating each candidate label equally or iteratively
identifying the true label. Nonetheless, existing algorithms usually treat all
labels and instances equally, and the complexities of both labels and instances
are not taken into consideration during the learning stage. Inspired by the
successful application of self-paced learning strategy in machine learning
field, we integrate the self-paced regime into the partial label learning
framework and propose a novel Self-Paced Partial-Label Learning (SP-PLL)
algorithm, which could control the learning process to alleviate the problem by
ranking the priorities of the training examples together with their candidate
labels during each learning iteration. Extensive experiments and comparisons
with other baseline methods demonstrate the effectiveness and robustness of the
proposed method.
|
Sampling the Riemann-Theta Boltzmann Machine | We show that the visible sector probability density function of the
Riemann-Theta Boltzmann machine corresponds to a gaussian mixture model
consisting of an infinite number of component multi-variate gaussians. The
weights of the mixture are given by a discrete multi-variate gaussian over the
hidden state space. This allows us to sample the visible sector density
function in a straight-forward manner. Furthermore, we show that the visible
sector probability density function possesses an affine transform property,
similar to the multi-variate gaussian density.
|
PEORL: Integrating Symbolic Planning and Hierarchical Reinforcement
Learning for Robust Decision-Making | Reinforcement learning and symbolic planning have both been used to build
intelligent autonomous agents. Reinforcement learning relies on learning from
interactions with real world, which often requires an unfeasibly large amount
of experience. Symbolic planning relies on manually crafted symbolic knowledge,
which may not be robust to domain uncertainties and changes. In this paper we
present a unified framework {\em PEORL} that integrates symbolic planning with
hierarchical reinforcement learning (HRL) to cope with decision-making in a
dynamic environment with uncertainties.
Symbolic plans are used to guide the agent's task execution and learning, and
the learned experience is fed back to symbolic knowledge to improve planning.
This method leads to rapid policy search and robust symbolic plans in complex
domains. The framework is tested on benchmark domains of HRL.
|
Generating Descriptions from Structured Data Using a Bifocal Attention
Mechanism and Gated Orthogonalization | In this work, we focus on the task of generating natural language
descriptions from a structured table of facts containing fields (such as
nationality, occupation, etc) and values (such as Indian, actor, director,
etc). One simple choice is to treat the table as a sequence of fields and
values and then use a standard seq2seq model for this task. However, such a
model is too generic and does not exploit task-specific characteristics. For
example, while generating descriptions from a table, a human would attend to
information at two levels: (i) the fields (macro level) and (ii) the values
within the field (micro level). Further, a human would continue attending to a
field for a few timesteps till all the information from that field has been
rendered and then never return back to this field (because there is nothing
left to say about it). To capture this behavior we use (i) a fused bifocal
attention mechanism which exploits and combines this micro and macro level
information and (ii) a gated orthogonalization mechanism which tries to ensure
that a field is remembered for a few time steps and then forgotten. We
experiment with a recently released dataset which contains fact tables about
people and their corresponding one line biographical descriptions in English.
In addition, we also introduce two similar datasets for French and German. Our
experiments show that the proposed model gives 21% relative improvement over a
recently proposed state of the art method and 10% relative improvement over
basic seq2seq models. The code and the datasets developed as a part of this
work are publicly available.
|
Stochastic subgradient method converges on tame functions | This work considers the question: what convergence guarantees does the
stochastic subgradient method have in the absence of smoothness and convexity?
We prove that the stochastic subgradient method, on any semialgebraic locally
Lipschitz function, produces limit points that are all first-order stationary.
More generally, our result applies to any function with a Whitney stratifiable
graph. In particular, this work endows the stochastic subgradient method, and
its proximal extension, with rigorous convergence guarantees for a wide class
of problems arising in data science---including all popular deep learning
architectures.
|
Value-aware Quantization for Training and Inference of Neural Networks | We propose a novel value-aware quantization which applies aggressively
reduced precision to the majority of data while separately handling a small
amount of large data in high precision, which reduces total quantization errors
under very low precision. We present new techniques to apply the proposed
quantization to training and inference. The experiments show that our method
with 3-bit activations (with 2% of large ones) can give the same training
accuracy as full-precision one while offering significant (41.6% and 53.7%)
reductions in the memory cost of activations in ResNet-152 and Inception-v3
compared with the state-of-the-art method. Our experiments also show that deep
networks such as Inception-v3, ResNet-101 and DenseNet-121 can be quantized for
inference with 4-bit weights and activations (with 1% 16-bit data) within 1%
top-1 accuracy drop.
|
Autotune: A Derivative-free Optimization Framework for Hyperparameter
Tuning | Machine learning applications often require hyperparameter tuning. The
hyperparameters usually drive both the efficiency of the model training process
and the resulting model quality. For hyperparameter tuning, machine learning
algorithms are complex black-boxes. This creates a class of challenging
optimization problems, whose objective functions tend to be nonsmooth,
discontinuous, unpredictably varying in computational expense, and include
continuous, categorical, and/or integer variables. Further, function
evaluations can fail for a variety of reasons including numerical difficulties
or hardware failures. Additionally, not all hyperparameter value combinations
are compatible, which creates so called hidden constraints. Robust and
efficient optimization algorithms are needed for hyperparameter tuning. In this
paper we present an automated parallel derivative-free optimization framework
called \textbf{Autotune}, which combines a number of specialized sampling and
search methods that are very effective in tuning machine learning models
despite these challenges. Autotune provides significantly improved models over
using default hyperparameter settings with minimal user interaction on
real-world applications. Given the inherent expense of training numerous
candidate models, we demonstrate the effectiveness of Autotune's search methods
and the efficient distributed and parallel paradigms for training and tuning
models, and also discuss the resource trade-offs associated with the ability to
both distribute the training process and parallelize the tuning process.
|
Online Improper Learning with an Approximation Oracle | We revisit the question of reducing online learning to approximate
optimization of the offline problem. In this setting, we give two algorithms
with near-optimal performance in the full information setting: they guarantee
optimal regret and require only poly-logarithmically many calls to the
approximation oracle per iteration. Furthermore, these algorithms apply to the
more general improper learning problems. In the bandit setting, our algorithm
also significantly improves the best previously known oracle complexity while
maintaining the same regret.
|
CactusNets: Layer Applicability as a Metric for Transfer Learning | Deep neural networks trained over large datasets learn features that are both
generic to the whole dataset, and specific to individual classes in the
dataset. Learned features tend towards generic in the lower layers and specific
in the higher layers of a network. Methods like fine-tuning are made possible
because of the ability for one filter to apply to multiple target classes. Much
like the human brain this behavior, can also be used to cluster and separate
classes. However, to the best of our knowledge there is no metric for how
applicable learned features are to specific classes. In this paper we propose a
definition and metric for measuring the applicability of learned features to
individual classes, and use this applicability metric to estimate input
applicability and produce a new method of unsupervised learning we call the
CactusNet.
|
Subgoal Discovery for Hierarchical Dialogue Policy Learning | Developing agents to engage in complex goal-oriented dialogues is challenging
partly because the main learning signals are very sparse in long conversations.
In this paper, we propose a divide-and-conquer approach that discovers and
exploits the hidden structure of the task to enable efficient policy learning.
First, given successful example dialogues, we propose the Subgoal Discovery
Network (SDN) to divide a complex goal-oriented task into a set of simpler
subgoals in an unsupervised fashion. We then use these subgoals to learn a
multi-level policy by hierarchical reinforcement learning. We demonstrate our
method by building a dialogue agent for the composite task of travel planning.
Experiments with simulated and real users show that our approach performs
competitively against a state-of-the-art method that requires human-defined
subgoals. Moreover, we show that the learned subgoals are often human
comprehensible.
|
Gradient Masking Causes CLEVER to Overestimate Adversarial Perturbation
Size | A key problem in research on adversarial examples is that vulnerability to
adversarial examples is usually measured by running attack algorithms. Because
the attack algorithms are not optimal, the attack algorithms are prone to
overestimating the size of perturbation needed to fool the target model. In
other words, the attack-based methodology provides an upper-bound on the size
of a perturbation that will fool the model, but security guarantees require a
lower bound. CLEVER is a proposed scoring method to estimate a lower bound.
Unfortunately, an estimate of a bound is not a bound. In this report, we show
that gradient masking, a common problem that causes attack methodologies to
provide only a very loose upper bound, causes CLEVER to overestimate the size
of perturbation needed to fool the model. In other words, CLEVER does not
resolve the key problem with the attack-based methodology, because it fails to
provide a lower bound.
|
Dynamic Ensemble Selection VS K-NN: why and when Dynamic Selection
obtains higher classification performance? | Multiple classifier systems focus on the combination of classifiers to obtain
better performance than a single robust one. These systems unfold three major
phases: pool generation, selection and integration. One of the most promising
MCS approaches is Dynamic Selection (DS), which relies on finding the most
competent classifier or ensemble of classifiers to predict each test sample.
The majority of the DS techniques are based on the K-Nearest Neighbors (K-NN)
definition, and the quality of the neighborhood has a huge impact on the
performance of DS methods. In this paper, we perform an analysis comparing the
classification results of DS techniques and the K-NN classifier under different
conditions. Experiments are performed on 18 state-of-the-art DS techniques over
30 classification datasets and results show that DS methods present a
significant boost in classification accuracy even though they use the same
neighborhood as the K-NN. The reasons behind the outperformance of DS
techniques over the K-NN classifier reside in the fact that DS techniques can
deal with samples with a high degree of instance hardness (samples that are
located close to the decision border) as opposed to the K-NN. In this paper,
not only we explain why DS techniques achieve higher classification performance
than the K-NN but also when DS should be used.
|
A Deep Learning Approach for Forecasting Air Pollution in South Korea
Using LSTM | Tackling air pollution is an imperative problem in South Korea, especially in
urban areas, over the last few years. More specially, South Korea has joined
the ranks of the world's most polluted countries alongside with other Asian
capitals, such as Beijing or Delhi. Much research is being conducted in
environmental science to evaluate the dangerous impact of particulate matters
on public health. Besides that, deterministic models of air pollutant behavior
are also generated; however, this is both complex and often inaccurate. On the
contrary, deep recurrent neural network reveals potent potential on forecasting
out-comes of time-series data and has become more prevalent. This paper uses
Recurrent Neural Network (RNN) with Long Short-Term Memory units as a framework
for leveraging knowledge from time-series data of air pollution and
meteorological information in Daegu, Seoul, Beijing, and Shenyang.
Additionally, we use encoder-decoder model, which is similar to machine
comprehension problems, as a crucial part of our prediction machine. Finally,
we investigate the prediction accuracy of various configurations. Our
experiments prevent the efficiency of integrating multiple layers of RNN on
prediction model when forecasting far timesteps ahead. This research is a
significant motivation for not only continuing researching on urban air quality
but also help the government leverage that insight to enact beneficial policies
|
Entire Space Multi-Task Model: An Effective Approach for Estimating
Post-Click Conversion Rate | Estimating post-click conversion rate (CVR) accurately is crucial for ranking
systems in industrial applications such as recommendation and advertising.
Conventional CVR modeling applies popular deep learning methods and achieves
state-of-the-art performance. However it encounters several task-specific
problems in practice, making CVR modeling challenging. For example,
conventional CVR models are trained with samples of clicked impressions while
utilized to make inference on the entire space with samples of all impressions.
This causes a sample selection bias problem. Besides, there exists an extreme
data sparsity problem, making the model fitting rather difficult. In this
paper, we model CVR in a brand-new perspective by making good use of sequential
pattern of user actions, i.e., impression -> click -> conversion. The proposed
Entire Space Multi-task Model (ESMM) can eliminate the two problems
simultaneously by i) modeling CVR directly over the entire space, ii) employing
a feature representation transfer learning strategy. Experiments on dataset
gathered from Taobao's recommender system demonstrate that ESMM significantly
outperforms competitive methods. We also release a sampling version of this
dataset to enable future research. To the best of our knowledge, this is the
first public dataset which contains samples with sequential dependence of click
and conversion labels for CVR modeling.
|
Is feature selection secure against training data poisoning? | Learning in adversarial settings is becoming an important task for
application domains where attackers may inject malicious data into the training
set to subvert normal operation of data-driven technologies. Feature selection
has been widely used in machine learning for security applications to improve
generalization and computational efficiency, although it is not clear whether
its use may be beneficial or even counterproductive when training data are
poisoned by intelligent attackers. In this work, we shed light on this issue by
providing a framework to investigate the robustness of popular feature
selection methods, including LASSO, ridge regression and the elastic net. Our
results on malware detection show that feature selection methods can be
significantly compromised under attack (we can reduce LASSO to almost random
choices of feature sets by careful insertion of less than 5% poisoned training
samples), highlighting the need for specific countermeasures.
|
Variational Inference In Pachinko Allocation Machines | The Pachinko Allocation Machine (PAM) is a deep topic model that allows
representing rich correlation structures among topics by a directed acyclic
graph over topics. Because of the flexibility of the model, however,
approximate inference is very difficult. Perhaps for this reason, only a small
number of potential PAM architectures have been explored in the literature. In
this paper we present an efficient and flexible amortized variational inference
method for PAM, using a deep inference network to parameterize the approximate
posterior distribution in a manner similar to the variational autoencoder. Our
inference method produces more coherent topics than state-of-art inference
methods for PAM while being an order of magnitude faster, which allows
exploration of a wider range of PAM architectures than have previously been
studied.
|
Stability of the Stochastic Gradient Method for an Approximated Large
Scale Kernel Machine | In this paper we measured the stability of stochastic gradient method (SGM)
for learning an approximated Fourier primal support vector machine. The
stability of an algorithm is considered by measuring the generalization error
in terms of the absolute difference between the test and the training error.
Our problem is to learn an approximated kernel function using random Fourier
features for a binary classification problem via online convex optimization
settings. For a convex, Lipschitz continuous and smooth loss function, given
reasonable number of iterations stochastic gradient method is stable. We showed
that with a high probability SGM generalizes well for an approximated kernel
under given assumptions.We empirically verified the theoretical findings for
different parameters using several data sets.
|
MQGrad: Reinforcement Learning of Gradient Quantization in Parameter
Server | One of the most significant bottleneck in training large scale machine
learning models on parameter server (PS) is the communication overhead, because
it needs to frequently exchange the model gradients between the workers and
servers during the training iterations. Gradient quantization has been proposed
as an effective approach to reducing the communication volume. One key issue in
gradient quantization is setting the number of bits for quantizing the
gradients. Small number of bits can significantly reduce the communication
overhead while hurts the gradient accuracies, and vise versa. An ideal
quantization method would dynamically balance the communication overhead and
model accuracy, through adjusting the number bits according to the knowledge
learned from the immediate past training iterations. Existing methods, however,
quantize the gradients either with fixed number of bits, or with predefined
heuristic rules. In this paper we propose a novel adaptive quantization method
within the framework of reinforcement learning. The method, referred to as
MQGrad, formalizes the selection of quantization bits as actions in a Markov
decision process (MDP) where the MDP states records the information collected
from the past optimization iterations (e.g., the sequence of the loss function
values). During the training iterations of a machine learning algorithm, MQGrad
continuously updates the MDP state according to the changes of the loss
function. Based on the information, MDP learns to select the optimal actions
(number of bits) to quantize the gradients. Experimental results based on a
benchmark dataset showed that MQGrad can accelerate the learning of a large
scale deep neural network while keeping its prediction accuracies.
|
Decoupled Networks | Inner product-based convolution has been a central component of convolutional
neural networks (CNNs) and the key to learning visual representations. Inspired
by the observation that CNN-learned features are naturally decoupled with the
norm of features corresponding to the intra-class variation and the angle
corresponding to the semantic difference, we propose a generic decoupled
learning framework which models the intra-class variation and semantic
difference independently. Specifically, we first reparametrize the inner
product to a decoupled form and then generalize it to the decoupled convolution
operator which serves as the building block of our decoupled networks. We
present several effective instances of the decoupled convolution operator. Each
decoupled operator is well motivated and has an intuitive geometric
interpretation. Based on these decoupled operators, we further propose to
directly learn the operator from data. Extensive experiments show that such
decoupled reparameterization renders significant performance gain with easier
convergence and stronger robustness.
|
Sparse Travel Time Estimation from Streaming Data | We address two shortcomings in online travel time estimation methods for
congested urban traffic. The first shortcoming is related to the determination
of the number of mixture modes, which can change dynamically, within day and
from day to day. The second shortcoming is the wide-spread use of Gaussian
probability densities as mixture components. Gaussian densities fail to capture
the positive skew in travel time distributions and, consequently, large numbers
of mixture components are needed for reasonable fitting accuracy when applied
as mixture components. They also assign positive probabilities to negative
travel times. To address these issues, this paper derives a mixture
distribution with Gamma component densities, which are asymmetric and supported
on the positive numbers. We use sparse estimation techniques to ensure
parsimonious models and propose a generalization of Gamma mixture densities
using Mittag-Leffler functions, which provides enhanced fitting flexibility and
improved parsimony. In order to accommodate within-day variability and allow
for online implementation of the proposed methodology (i.e., fast computations
on streaming travel time data), we introduce a recursive algorithm which
efficiently updates the fitted distribution whenever new data become available.
Experimental results using real-world travel time data illustrate the efficacy
of the proposed methods.
|
Adaptive Performance Assessment For Drivers Through Behavioral Advantage | The potential positive impact of autonomous driving and driver assistance
technolo- gies have been a major impetus over the last decade. On the flip
side, it has been a challenging problem to analyze the performance of human
drivers or autonomous driving agents quantitatively. In this work, we propose a
generic method that compares the performance of drivers or autonomous driving
agents even if the environmental conditions are different, by using the driver
behavioral advantage instead of absolute metrics, which efficiently removes the
environmental factors. A concrete application of the method is also presented,
where the performance of more than 100 truck drivers was evaluated and ranked
in terms of fuel efficiency, covering more than 90,000 trips spanning an
average of 300 miles in a variety of driving conditions and environments.
|
N-fold Superposition: Improving Neural Networks by Reducing the Noise in
Feature Maps | Considering the use of Fully Connected (FC) layer limits the performance of
Convolutional Neural Networks (CNNs), this paper develops a method to improve
the coupling between the convolution layer and the FC layer by reducing the
noise in Feature Maps (FMs). Our approach is divided into three steps. Firstly,
we separate all the FMs into n blocks equally. Then, the weighted summation of
FMs at the same position in all blocks constitutes a new block of FMs. Finally,
we replicate this new block into n copies and concatenate them as the input to
the FC layer. This sharing of FMs could reduce the noise in them apparently and
avert the impact by a particular FM on the specific part weight of hidden
layers, hence preventing the network from overfitting to some extent. Using the
Fermat Lemma, we prove that this method could make the global minima value
range of the loss function wider, by which makes it easier for neural networks
to converge and accelerates the convergence process. This method does not
significantly increase the amounts of network parameters (only a few more
coefficients added), and the experiments demonstrate that this method could
increase the convergence speed and improve the classification performance of
neural networks.
|
Clinical Assistant Diagnosis for Electronic Medical Record Based on
Convolutional Neural Network | Automatically extracting useful information from electronic medical records
along with conducting disease diagnoses is a promising task for both clinical
decision support(CDS) and neural language processing(NLP). Most of the existing
systems are based on artificially constructed knowledge bases, and then
auxiliary diagnosis is done by rule matching. In this study, we present a
clinical intelligent decision approach based on Convolutional Neural
Networks(CNN), which can automatically extract high-level semantic information
of electronic medical records and then perform automatic diagnosis without
artificial construction of rules or knowledge bases. We use collected 18,590
copies of the real-world clinical electronic medical records to train and test
the proposed model. Experimental results show that the proposed model can
achieve 98.67\% accuracy and 96.02\% recall, which strongly supports that using
convolutional neural network to automatically learn high-level semantic
features of electronic medical records and then conduct assist diagnosis is
feasible and effective.
|
Deep cross-domain building extraction for selective depth estimation
from oblique aerial imagery | With the technological advancements of aerial imagery and accurate 3d
reconstruction of urban environments, more and more attention has been paid to
the automated analyses of urban areas. In our work, we examine two important
aspects that allow live analysis of building structures in city models given
oblique aerial imagery, namely automatic building extraction with convolutional
neural networks (CNNs) and selective real-time depth estimation from aerial
imagery. We use transfer learning to train the Faster R-CNN method for
real-time deep object detection, by combining a large ground-based dataset for
urban scene understanding with a smaller number of images from an aerial
dataset. We achieve an average precision (AP) of about 80% for the task of
building extraction on a selected evaluation dataset. Our evaluation focuses on
both dataset-specific learning and transfer learning. Furthermore, we present
an algorithm that allows for multi-view depth estimation from aerial imagery in
real-time. We adopt the semi-global matching (SGM) optimization strategy to
preserve sharp edges at object boundaries. In combination with the Faster
R-CNN, it allows a selective reconstruction of buildings, identified with
regions of interest (RoIs), from oblique aerial imagery.
|
Taskonomy: Disentangling Task Transfer Learning | Do visual tasks have a relationship, or are they unrelated? For instance,
could having surface normals simplify estimating the depth of an image?
Intuition answers these questions positively, implying existence of a structure
among visual tasks. Knowing this structure has notable values; it is the
concept underlying transfer learning and provides a principled way for
identifying redundancies across tasks, e.g., to seamlessly reuse supervision
among related tasks or solve many tasks in one system without piling up the
complexity.
We proposes a fully computational approach for modeling the structure of
space of visual tasks. This is done via finding (first and higher-order)
transfer learning dependencies across a dictionary of twenty six 2D, 2.5D, 3D,
and semantic tasks in a latent space. The product is a computational taxonomic
map for task transfer learning. We study the consequences of this structure,
e.g. nontrivial emerged relationships, and exploit them to reduce the demand
for labeled data. For example, we show that the total number of labeled
datapoints needed for solving a set of 10 tasks can be reduced by roughly 2/3
(compared to training independently) while keeping the performance nearly the
same. We provide a set of tools for computing and probing this taxonomical
structure including a solver that users can employ to devise efficient
supervision policies for their use cases.
|
Client Selection for Federated Learning with Heterogeneous Resources in
Mobile Edge | We envision a mobile edge computing (MEC) framework for machine learning (ML)
technologies, which leverages distributed client data and computation resources
for training high-performance ML models while preserving client privacy. Toward
this future goal, this work aims to extend Federated Learning (FL), a
decentralized learning framework that enables privacy-preserving training of
models, to work with heterogeneous clients in a practical cellular network. The
FL protocol iteratively asks random clients to download a trainable model from
a server, update it with own data, and upload the updated model to the server,
while asking the server to aggregate multiple client updates to further improve
the model. While clients in this protocol are free from disclosing own private
data, the overall training process can become inefficient when some clients are
with limited computational resources (i.e. requiring longer update time) or
under poor wireless channel conditions (longer upload time). Our new FL
protocol, which we refer to as FedCS, mitigates this problem and performs FL
efficiently while actively managing clients based on their resource conditions.
Specifically, FedCS solves a client selection problem with resource
constraints, which allows the server to aggregate as many client updates as
possible and to accelerate performance improvement in ML models. We conducted
an experimental evaluation using publicly-available large-scale image datasets
to train deep neural networks on MEC environment simulations. The experimental
results show that FedCS is able to complete its training process in a
significantly shorter time compared to the original FL protocol.
|
Gaussian Material Synthesis | We present a learning-based system for rapid mass-scale material synthesis
that is useful for novice and expert users alike. The user preferences are
learned via Gaussian Process Regression and can be easily sampled for new
recommendations. Typically, each recommendation takes 40-60 seconds to render
with global illumination, which makes this process impracticable for real-world
workflows. Our neural network eliminates this bottleneck by providing
high-quality image predictions in real time, after which it is possible to pick
the desired materials from a gallery and assign them to a scene in an intuitive
manner. Workflow timings against Disney's "principled" shader reveal that our
system scales well with the number of sought materials, thus empowering even
novice users to generate hundreds of high-quality material models without any
expertise in material modeling. Similarly, expert users experience a
significant decrease in the total modeling time when populating a scene with
materials. Furthermore, our proposed solution also offers controllable
recommendations and a novel latent space variant generation step to enable the
real-time fine-tuning of materials without requiring any domain expertise.
|
Convolutional capsule network for classification of breast cancer
histology images | Automatization of the diagnosis of any kind of disease is of great importance
and it's gaining speed as more and more deep learning solutions are applied to
different problems. One of such computer aided systems could be a decision
support too able to accurately differentiate between different types of breast
cancer histological images - normal tissue or carcinoma. In this paper authors
present a deep learning solution, based on convolutional capsule network for
classification of four types of images of breast tissue biopsy when hematoxylin
and eusin staining is applied. The cross-validation accuracy was achieved to be
0.87 with equaly high sensitivity.
|
Path Planning in Support of Smart Mobility Applications using Generative
Adversarial Networks | This paper describes and evaluates the use of Generative Adversarial Networks
(GANs) for path planning in support of smart mobility applications such as
indoor and outdoor navigation applications, individualized wayfinding for
people with disabilities (e.g., vision impairments, physical disabilities,
etc.), path planning for evacuations, robotic navigations, and path planning
for autonomous vehicles. We propose an architecture based on GANs to recommend
accurate and reliable paths for navigation applications. The proposed system
can use crowd-sourced data to learn the trajectories and infer new ones. The
system provides users with generated paths that help them navigate from their
local environment to reach a desired location. As a use case, we experimented
with the proposed method in support of a wayfinding application in an indoor
environment. Our experiments assert that the generated paths are correct and
reliable. The accuracy of the classification task for the generated paths is up
to 99% and the quality of the generated paths has a mean opinion score of 89%.
|
Learn and Pick Right Nodes to Offload | Task offloading is a promising technology to exploit the benefits of fog
computing. An effective task offloading strategy is needed to utilize the
computational resources efficiently. In this paper, we endeavor to seek an
online task offloading strategy to minimize the long-term latency. In
particular, we formulate a stochastic programming problem, where the
expectations of the system parameters change abruptly at unknown time instants.
Meanwhile, we consider the fact that the queried nodes can only feed back the
processing results after finishing the tasks. We then put forward an effective
algorithm to solve this challenging stochastic programming under the
non-stationary bandit model. We further prove that our proposed algorithm is
asymptotically optimal in a non-stationary fog-enabled network. Numerical
simulations are carried out to corroborate our designs.
|
Exploiting Partially Annotated Data for Temporal Relation Extraction | Annotating temporal relations (TempRel) between events described in natural
language is known to be labor intensive, partly because the total number of
TempRels is quadratic in the number of events. As a result, only a small number
of documents are typically annotated, limiting the coverage of various
lexical/semantic phenomena. In order to improve existing approaches, one
possibility is to make use of the readily available, partially annotated data
(P as in partial) that cover more documents. However, missing annotations in P
are known to hurt, rather than help, existing systems. This work is a case
study in exploring various usages of P for TempRel extraction. Results show
that despite missing annotations, P is still a useful supervision signal for
this task within a constrained bootstrapping learning framework. The system
described in this system is publicly available.
|
Decorrelated Batch Normalization | Batch Normalization (BN) is capable of accelerating the training of deep
models by centering and scaling activations within mini-batches. In this work,
we propose Decorrelated Batch Normalization (DBN), which not just centers and
scales activations but whitens them. We explore multiple whitening techniques,
and find that PCA whitening causes a problem we call stochastic axis swapping,
which is detrimental to learning. We show that ZCA whitening does not suffer
from this problem, permitting successful learning. DBN retains the desirable
qualities of BN and further improves BN's optimization efficiency and
generalization ability. We design comprehensive experiments to show that DBN
can improve the performance of BN on multilayer perceptrons and convolutional
neural networks. Furthermore, we consistently improve the accuracy of residual
networks on CIFAR-10, CIFAR-100, and ImageNet.
|
Attention Based Natural Language Grounding by Navigating Virtual
Environment | In this work, we focus on the problem of grounding language by training an
agent to follow a set of natural language instructions and navigate to a target
object in an environment. The agent receives visual information through raw
pixels and a natural language instruction telling what task needs to be
achieved and is trained in an end-to-end way. We develop an attention mechanism
for multi-modal fusion of visual and textual modalities that allows the agent
to learn to complete the task and achieve language grounding. Our experimental
results show that our attention mechanism outperforms the existing multi-modal
fusion mechanisms proposed for both 2D and 3D environments in order to solve
the above-mentioned task in terms of both speed and success rate. We show that
the learnt textual representations are semantically meaningful as they follow
vector arithmetic in the embedding space. The effectiveness of our attention
approach over the contemporary fusion mechanisms is also highlighted from the
textual embeddings learnt by the different approaches. We also show that our
model generalizes effectively to unseen scenarios and exhibit zero-shot
generalization capabilities both in 2D and 3D environments. The code for our 2D
environment as well as the models that we developed for both 2D and 3D are
available at https://github.com/rl-lang-grounding/rl-lang-ground.
|
ALIGNet: Partial-Shape Agnostic Alignment via Unsupervised Learning | The process of aligning a pair of shapes is a fundamental operation in
computer graphics. Traditional approaches rely heavily on matching
corresponding points or features to guide the alignment, a paradigm that
falters when significant shape portions are missing. These techniques generally
do not incorporate prior knowledge about expected shape characteristics, which
can help compensate for any misleading cues left by inaccuracies exhibited in
the input shapes. We present an approach based on a deep neural network,
leveraging shape datasets to learn a shape-aware prior for source-to-target
alignment that is robust to shape incompleteness. In the absence of ground
truth alignments for supervision, we train a network on the task of shape
alignment using incomplete shapes generated from full shapes for
self-supervision. Our network, called ALIGNet, is trained to warp complete
source shapes to incomplete targets, as if the target shapes were complete,
thus essentially rendering the alignment partial-shape agnostic. We aim for the
network to develop specialized expertise over the common characteristics of the
shapes in each dataset, thereby achieving a higher-level understanding of the
expected shape space to which a local approach would be oblivious. We constrain
ALIGNet through an anisotropic total variation identity regularization to
promote piecewise smooth deformation fields, facilitating both partial-shape
agnosticism and post-deformation applications. We demonstrate that ALIGNet
learns to align geometrically distinct shapes, and is able to infer plausible
mappings even when the target shape is significantly incomplete. We show that
our network learns the common expected characteristics of shape collections,
without over-fitting or memorization, enabling it to produce plausible
deformations on unseen data during test time.
|
Dropping Networks for Transfer Learning | Many tasks in natural language understanding require learning relationships
between two sequences for various tasks such as natural language inference,
paraphrasing and entailment. These aforementioned tasks are similar in nature,
yet they are often modeled individually. Knowledge transfer can be effective
for closely related tasks. However, transferring all knowledge, some of which
irrelevant for a target task, can lead to sub-optimal results due to
\textit{negative} transfer. Hence, this paper focuses on the transferability of
both instances and parameters across natural language understanding tasks by
proposing an ensemble-based transfer learning method. \newline The primary
contribution of this paper is the combination of both \textit{Dropout} and
\textit{Bagging} for improved transferability in neural networks, referred to
as \textit{Dropping} herein. We present a straightforward yet novel approach
for incorporating source \textit{Dropping} Networks to a target task for
few-shot learning that mitigates \textit{negative} transfer. This is achieved
by using a decaying parameter chosen according to the slope changes of a
smoothed spline error curve at sub-intervals during training. We compare the
proposed approach against hard parameter sharing and soft parameter sharing
transfer methods in the few-shot learning case. We also compare against models
that are fully trained on the target task in the standard supervised learning
setup. The aforementioned adjustment leads to improved transfer learning
performance and comparable results to the current state of the art only using a
fraction of the data from the target task.
|
VectorDefense: Vectorization as a Defense to Adversarial Examples | Training deep neural networks on images represented as grids of pixels has
brought to light an interesting phenomenon known as adversarial examples.
Inspired by how humans reconstruct abstract concepts, we attempt to codify the
input bitmap image into a set of compact, interpretable elements to avoid being
fooled by the adversarial structures. We take the first step in this direction
by experimenting with image vectorization as an input transformation step to
map the adversarial examples back into the natural manifold of MNIST
handwritten digits. We compare our method vs. state-of-the-art input
transformations and further discuss the trade-offs between a hand-designed and
a learned transformation defense.
|
Spatio-Temporal Neural Networks for Space-Time Series Forecasting and
Relations Discovery | We introduce a dynamical spatio-temporal model formalized as a recurrent
neural network for forecasting time series of spatial processes, i.e. series of
observations sharing temporal and spatial dependencies. The model learns these
dependencies through a structured latent dynamical component, while a decoder
predicts the observations from the latent representations. We consider several
variants of this model, corresponding to different prior hypothesis about the
spatial relations between the series. The model is evaluated and compared to
state-of-the-art baselines, on a variety of forecasting problems representative
of different application areas: epidemiology, geo-spatial statistics and
car-traffic prediction. Besides these evaluations, we also describe experiments
showing the ability of this approach to extract relevant spatial relations.
|
Towards Symbolic Reinforcement Learning with Common Sense | Deep Reinforcement Learning (deep RL) has made several breakthroughs in
recent years in applications ranging from complex control tasks in unmanned
vehicles to game playing. Despite their success, deep RL still lacks several
important capacities of human intelligence, such as transfer learning,
abstraction and interpretability. Deep Symbolic Reinforcement Learning (DSRL)
seeks to incorporate such capacities to deep Q-networks (DQN) by learning a
relevant symbolic representation prior to using Q-learning. In this paper, we
propose a novel extension of DSRL, which we call Symbolic Reinforcement
Learning with Common Sense (SRL+CS), offering a better balance between
generalization and specialization, inspired by principles of common sense when
assigning rewards and aggregating Q-values. Experiments reported in this paper
show that SRL+CS learns consistently faster than Q-learning and DSRL, achieving
also a higher accuracy. In the hardest case, where agents were trained in a
deterministic environment and tested in a random environment, SRL+CS achieves
nearly 100% average accuracy compared to DSRL's 70% and DQN's 50% accuracy. To
the best of our knowledge, this is the first case of near perfect zero-shot
transfer learning using Reinforcement Learning.
|
Towards Learning Sparsely Used Dictionaries with Arbitrary Supports | Dictionary learning is a popular approach for inferring a hidden basis or
dictionary in which data has a sparse representation. Data generated from the
dictionary A (an n by m matrix, with m > n in the over-complete setting) is
given by Y = AX where X is a matrix whose columns have supports chosen from a
distribution over k-sparse vectors, and the non-zero values chosen from a
symmetric distribution. Given Y, the goal is to recover A and X in polynomial
time. Existing algorithms give polytime guarantees for recovering incoherent
dictionaries, under strong distributional assumptions both on the supports of
the columns of X, and on the values of the non-zero entries. In this work, we
study the following question: Can we design efficient algorithms for recovering
dictionaries when the supports of the columns of X are arbitrary?
To address this question while circumventing the issue of
non-identifiability, we study a natural semirandom model for dictionary
learning where there are a large number of samples $y=Ax$ with arbitrary
k-sparse supports for x, along with a few samples where the sparse supports are
chosen uniformly at random. While the few samples with random supports ensures
identifiability, the support distribution can look almost arbitrary in
aggregate. Hence existing algorithmic techniques seem to break down as they
make strong assumptions on the supports.
Our main contribution is a new polynomial time algorithm for learning
incoherent over-complete dictionaries that works under the semirandom model.
Additionally the same algorithm provides polynomial time guarantees in new
parameter regimes when the supports are fully random. Finally using these
techniques, we also identify a minimal set of conditions on the supports under
which the dictionary can be (information theoretically) recovered from
polynomial samples for almost linear sparsity, i.e., $k=\tilde{O}(n)$.
|
Zero-Shot Visual Imitation | The current dominant paradigm for imitation learning relies on strong
supervision of expert actions to learn both 'what' and 'how' to imitate. We
pursue an alternative paradigm wherein an agent first explores the world
without any expert supervision and then distills its experience into a
goal-conditioned skill policy with a novel forward consistency loss. In our
framework, the role of the expert is only to communicate the goals (i.e., what
to imitate) during inference. The learned policy is then employed to mimic the
expert (i.e., how to imitate) after seeing just a sequence of images
demonstrating the desired task. Our method is 'zero-shot' in the sense that the
agent never has access to expert actions during training or for the task
demonstration at inference. We evaluate our zero-shot imitator in two
real-world settings: complex rope manipulation with a Baxter robot and
navigation in previously unseen office environments with a TurtleBot. Through
further experiments in VizDoom simulation, we provide evidence that better
mechanisms for exploration lead to learning a more capable policy which in turn
improves end task performance. Videos, models, and more details are available
at https://pathak22.github.io/zeroshot-imitation/
|
Benchmarking projective simulation in navigation problems | Projective simulation (PS) is a model for intelligent agents with a
deliberation capacity that is based on episodic memory. The model has been
shown to provide a flexible framework for constructing reinforcement-learning
agents, and it allows for quantum mechanical generalization, which leads to a
speed-up in deliberation time. PS agents have been applied successfully in the
context of complex skill learning in robotics, and in the design of
state-of-the-art quantum experiments. In this paper, we study the performance
of projective simulation in two benchmarking problems in navigation, namely the
grid world and the mountain car problem. The performance of PS is compared to
standard tabular reinforcement learning approaches, Q-learning and SARSA. Our
comparison demonstrates that the performance of PS and standard learning
approaches are qualitatively and quantitatively similar, while it is much
easier to choose optimal model parameters in case of projective simulation,
with a reduced computational effort of one to two orders of magnitude. Our
results show that the projective simulation model stands out for its simplicity
in terms of the number of model parameters, which makes it simple to set up the
learning agent in unknown task environments.
|
Parameter Transfer Unit for Deep Neural Networks | Parameters in deep neural networks which are trained on large-scale databases
can generalize across multiple domains, which is referred as "transferability".
Unfortunately, the transferability is usually defined as discrete states and it
differs with domains and network architectures. Existing works usually
heuristically apply parameter-sharing or fine-tuning, and there is no
principled approach to learn a parameter transfer strategy. To address the gap,
a parameter transfer unit (PTU) is proposed in this paper. The PTU learns a
fine-grained nonlinear combination of activations from both the source and the
target domain networks, and subsumes hand-crafted discrete transfer states. In
the PTU, the transferability is controlled by two gates which are artificial
neurons and can be learned from data. The PTU is a general and flexible module
which can be used in both CNNs and RNNs. Experiments are conducted with various
network architectures and multiple transfer domain pairs. Results demonstrate
the effectiveness of the PTU as it outperforms heuristic parameter-sharing and
fine-tuning in most settings.
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.