title
stringlengths 6
244
| abstract
stringlengths 19
5.09k
| label
stringclasses 10
values |
|---|---|---|
Optimal Transport Graph Neural Networks | Current graph neural network (GNN) architectures naively average or sum node
embeddings into an aggregated graph representation -- potentially losing
structural or semantic information. We here introduce OT-GNN, a model that
computes graph embeddings using parametric prototypes that highlight key facets
of different graph aspects. Towards this goal, we are (to our knowledge) the
first to successfully combine optimal transport (OT) with parametric graph
models. Graph representations are obtained from Wasserstein distances between
the set of GNN node embeddings and "prototype" point clouds as free parameters.
We theoretically prove that, unlike traditional sum aggregation, our function
class on point clouds satisfies a fundamental universal approximation theorem.
Empirically, we address an inherent collapse optimization issue by proposing a
noise contrastive regularizer to steer the model towards truly exploiting the
optimal transport geometry. Finally, we consistently report better
generalization performance on several molecular property prediction tasks,
while exhibiting smoother graph representations.
| stat |
COVID-19: Optimal Design of Serosurveys for Disease Burden Estimation | We provide a methodology by which an epidemiologist may arrive at an optimal
design for a survey whose goal is to estimate the disease burden in a
population. For serosurveys with a given budget of $C$ rupees, a specified set
of tests with costs, sensitivities, and specificities, we show the existence of
optimal designs in four different contexts, including the well known c-optimal
design. Usefulness of the results are illustrated via numerical examples. Our
results are applicable to a wide range of epidemiological surveys under the
assumptions that the estimate's Fisher-information matrix satisfies a uniform
positive definite criterion.
| stat |
Solar: a least-angle regression for stable variable selection in
high-dimensional spaces | We propose a new algorithm for variable selection in high-dimensional data,
called subsample-ordered least-angle regression (solar). Solar relies on the
average $L_0$ solution path computed across subsamples and alleviates several
known high-dimensional issues with lasso and least-angle regression. We
illustrate in simulations that, with the same computation load, solar yields
substantial improvements over lasso in terms of the sparsity (37-64\% reduction
in the average number of selected variables), stability and accuracy of
variable selection. Moreover, solar supplemented with the hold-out average (an
adaptation of classical post-OLS tests) successfully purges almost all of the
redundant variables while retaining all of the informative variables. Using
simulations and real-world data, we also illustrate numerically that sparse
solar variable selection is robust to complicated dependence structures and
harsh settings of the irrepresentable condition. Moreover, replacing lasso with
solar in an ensemble system (e.g., the bootstrap ensemble), significantly
reduces the computation load (at least 96\% fewer subsample repetitions) of the
bootstrap ensemble and improves selection sparsity. We provide a Python
parallel computing package for solar (solarpy) in the supplementary file and
https://github.com/isaac2math/solar.
| stat |
What causes the test error? Going beyond bias-variance via ANOVA | Modern machine learning methods are often overparametrized, allowing
adaptation to the data at a fine level. This can seem puzzling; in the worst
case, such models do not need to generalize. This puzzle inspired a great
amount of work, arguing when overparametrization reduces test error, in a
phenomenon called "double descent". Recent work aimed to understand in greater
depth why overparametrization is helpful for generalization. This leads to
discovering the unimodality of variance as a function of the level of
parametrization, and to decomposing the variance into that arising from label
noise, initialization, and randomness in the training data to understand the
sources of the error.
In this work we develop a deeper understanding of this area. Specifically, we
propose using the analysis of variance (ANOVA) to decompose the variance in the
test error in a symmetric way, for studying the generalization performance of
certain two-layer linear and non-linear networks. The advantage of the analysis
of variance is that it reveals the effects of initialization, label noise, and
training data more clearly than prior approaches. Moreover, we also study the
monotonicity and unimodality of the variance components. While prior work
studied the unimodality of the overall variance, we study the properties of
each term in variance decomposition.
One key insight is that in typical settings, the interaction between training
samples and initialization can dominate the variance; surprisingly being larger
than their marginal effect. Also, we characterize "phase transitions" where the
variance changes from unimodal to monotone. On a technical level, we leverage
advanced deterministic equivalent techniques for Haar random matrices,
that---to our knowledge---have not yet been used in the area. We also verify
our results in numerical simulations and on empirical data examples.
| stat |
Invited Discussion of "A Unified Framework for De-Duplication and
Population Size Estimation" | Invited Discussion of "A Unified Framework for De-Duplication and Population
Size Estimation", published in Bayesian Analysis. My discussion focuses on two
main themes: Providing a more nuanced picture of the costs and benefits of
joint models for record linkage and the "downstream task" (i.e. whatever we
might want to do with the linked and de-duplicated files), and how we should
measure performance.
| stat |
Dual Instrumental Variable Regression | We present a novel algorithm for non-linear instrumental variable (IV)
regression, DualIV, which simplifies traditional two-stage methods via a dual
formulation. Inspired by problems in stochastic programming, we show that
two-stage procedures for non-linear IV regression can be reformulated as a
convex-concave saddle-point problem. Our formulation enables us to circumvent
the first-stage regression which is a potential bottleneck in real-world
applications. We develop a simple kernel-based algorithm with an analytic
solution based on this formulation. Empirical results show that we are
competitive to existing, more complicated algorithms for non-linear
instrumental variable regression.
| stat |
Accelerated MM Algorithms for Ranking Scores Inference from Comparison
Data | In this paper, we study a popular method for inference of the Bradley-Terry
model parameters, namely the MM algorithm, for maximum likelihood estimation
and maximum a posteriori probability estimation. This class of models includes
the Bradley-Terry model of paired comparisons, the Rao-Kupper model of paired
comparisons allowing for tie outcomes, the Luce choice model, and the
Plackett-Luce ranking model. We establish tight characterizations of the
convergence rate for the MM algorithm, and show that it is essentially
equivalent to that of a gradient descent algorithm. For the maximum likelihood
estimation, the convergence is shown to be linear with the rate crucially
determined by the algebraic connectivity of the matrix of item pair
co-occurrences in observed comparison data. For the Bayesian inference, the
convergence rate is also shown to be linear, with the rate determined by a
parameter of the prior distribution in a way that can make the convergence
arbitrarily slow for small values of this parameter. We propose a simple
modification of the classical MM algorithm that avoids the observed slow
convergence issue and accelerates the convergence. The key component of the
accelerated MM algorithm is a parameter rescaling performed at each iteration
step that is carefully chosen based on theoretical analysis and
characterisation of the convergence rate.
Our experimental results, performed on both synthetic and real-world data,
demonstrate the identified slow convergence issue of the classic MM algorithm,
and show that significant efficiency gains can be obtained by our new proposed
method.
| stat |
A Flexible Procedure for Mixture Proportion Estimation in
Positive-Unlabeled Learning | Positive--unlabeled (PU) learning considers two samples, a positive set P
with observations from only one class and an unlabeled set U with observations
from two classes. The goal is to classify observations in U. Class mixture
proportion estimation (MPE) in U is a key step in PU learning. Blanchard et al.
[2010] showed that MPE in PU learning is a generalization of the problem of
estimating the proportion of true null hypotheses in multiple testing problems.
Motivated by this idea, we propose reducing the problem to one dimension via
construction of a probabilistic classifier trained on the P and U data sets
followed by application of a one--dimensional mixture proportion method from
the multiple testing literature to the observation class probabilities. The
flexibility of this framework lies in the freedom to choose the classifier and
the one--dimensional MPE method. We prove consistency of two mixture proportion
estimators using bounds from empirical process theory, develop tuning parameter
free implementations, and demonstrate that they have competitive performance on
simulated waveform data and a protein signaling problem.
| stat |
VarFA: A Variational Factor Analysis Framework For Efficient Bayesian
Learning Analytics | We propose VarFA, a variational inference factor analysis framework that
extends existing factor analysis models for educational data mining to
efficiently output uncertainty estimation in the model's estimated factors.
Such uncertainty information is useful, for example, for an adaptive testing
scenario, where additional tests can be administered if the model is not quite
certain about a students' skill level estimation. Traditional Bayesian
inference methods that produce such uncertainty information are computationally
expensive and do not scale to large data sets. VarFA utilizes variational
inference which makes it possible to efficiently perform Bayesian inference
even on very large data sets. We use the sparse factor analysis model as a case
study and demonstrate the efficacy of VarFA on both synthetic and real data
sets. VarFA is also very general and can be applied to a wide array of factor
analysis models.
| stat |
Rehabilitating Isomap: Euclidean Representation of Geodesic Structure | Manifold learning techniques for nonlinear dimension reduction assume that
high-dimensional feature vectors lie on a low-dimensional manifold, then
attempt to exploit manifold structure to obtain useful low-dimensional
Euclidean representations of the data. Isomap, a seminal manifold learning
technique, is an elegant synthesis of two simple ideas: the approximation of
Riemannian distances with shortest path distances on a graph that localizes
manifold structure, and the approximation of shortest path distances with
Euclidean distances by multidimensional scaling. We revisit the rationale for
Isomap, clarifying what Isomap does and what it does not. In particular, we
explore the widespread perception that Isomap should only be used when the
manifold is parametrized by a convex region of Euclidean space. We argue that
this perception is based on an extremely narrow interpretation of manifold
learning as parametrization recovery, and we submit that Isomap is better
understood as constructing Euclidean representations of geodesic structure. We
reconsider a well-known example that was previously interpreted as evidence of
Isomap's limitations, and we re-examine the original analysis of Isomap's
convergence properties, concluding that convexity is not required for shortest
path distances to converge to Riemannian distances.
| stat |
On Relative Ageing of Coherent Systems with Dependent Identically
Distributed Components | Relative ageing describes how a system ages with respect to another one. The
ageing faster orders are the ones which compare the relative ageings of two
systems. Here, we study ageing faster orders in the hazard and the reversed
hazard rates. We provide some sufficient conditions for proving that one
coherent system dominates another system with respect to ageing faster orders.
Further, we investigate whether the active redundancy at the component level is
more effective than that at the system level with respect to ageing faster
orders, for a coherent system. Furthermore, a used coherent system and a
coherent system made out of used components are compared with respect to ageing
faster orders.
| stat |
The implications of Labour's plan to scrap Key Stage 2 tests for
Progress 8 and secondary school accountability in England | In England, Progress 8 is the Conservative government's headline secondary
school performance and accountability measure. Progress 8 attempts to measure
the average academic progress pupils make in each school between their KS2
tests and their GCSE Attainment 8 examinations. The Labour opposition recently
announced they would scrap the KS2 tests were they to be elected. Such a move,
however, would preclude the publication of Progress 8 and would leave schools
to be compared in terms of their average Attainment 8 scores or, at best, their
Attainment 8 scores only adjusted for school differences in pupil demographic
and socioeconomic characteristics. In this paper, we argue and illustrate
empirically that this best-case scenario of an 'Adjusted Attainment 8' measure
would prove less fair and meaningful than Progress 8 and therefore a backwards
step, especially when Progress 8 itself has been criticised as biased against
schools teaching educationally disadvantaged intakes.
| stat |
Bayesian Kernel Two-Sample Testing | In modern data analysis, nonparametric measures of discrepancies between
random variables are particularly important. The subject is well-studied in the
frequentist literature, while the development in the Bayesian setting is
limited where applications are often restricted to univariate cases. Here, we
propose a Bayesian kernel two-sample testing procedure based on modelling the
difference between kernel mean embeddings in the reproducing kernel Hilbert
space utilising the framework established by Flaxman et al (2016). The use of
kernel methods enables its application to random variables in generic domains
beyond the multivariate Euclidean spaces. The proposed procedure results in a
posterior inference scheme that allows an automatic selection of the kernel
parameters relevant to the problem at hand. In a series of synthetic
experiments and two real data experiments (i.e. testing network heterogeneity
from high-dimensional data and six-membered monocyclic ring conformation
comparison), we illustrate the advantages of our approach.
| stat |
Efficient Selection Between Hierarchical Cognitive Models:
Cross-validation With Variational Bayes | Model comparison is the cornerstone of theoretical progress in psychological
research. Common practice overwhelmingly relies on tools that evaluate
competing models by balancing in-sample descriptive adequacy against model
flexibility, with modern approaches advocating the use of marginal likelihood
for hierarchical cognitive models. Cross-validation is another popular approach
but its implementation has remained out of reach for cognitive models evaluated
in a Bayesian hierarchical framework, with the major hurdle being prohibitive
computational cost. To address this issue, we develop novel algorithms that
make variational Bayes (VB) inference for hierarchical models feasible and
computationally efficient for complex cognitive models of substantive
theoretical interest. It is well known that VB produces good estimates of the
first moments of the parameters which gives good predictive densities
estimates. We thus develop a novel VB algorithm with Bayesian prediction as a
tool to perform model comparison by cross-validation, which we refer to as
CVVB. In particular, the CVVB can be used as a model screening device that
quickly identifies bad models. We demonstrate the utility of CVVB by revisiting
a classic question in decision making research: what latent components of
processing drive the ubiquitous speed-accuracy tradeoff? We demonstrate that
CVVB strongly agrees with model comparison via marginal likelihood yet achieves
the outcome in much less time. Our approach brings cross-validation within
reach of theoretically important psychological models, and makes it feasible to
compare much larger families of hierarchically specified cognitive models than
has previously been possible.
| stat |
Multivariate type G Mat\'ern stochastic partial differential equation
random fields | For many applications with multivariate data, random field models capturing
departures from Gaussianity within realisations are appropriate. For this
reason, we formulate a new class of multivariate non-Gaussian models based on
systems of stochastic partial differential equations with additive type G noise
whose marginal covariance functions are of Mat\'ern type. We consider four
increasingly flexible constructions of the noise, where the first two are
similar to existing copula-based models. In contrast to these, the latter two
constructions can model non-Gaussian spatial data without replicates.
Computationally efficient methods for likelihood-based parameter estimation and
probabilistic prediction are proposed, and the flexibility of the suggested
models is illustrated by numerical examples and two statistical applications.
| stat |
Learning Dynamical Systems using Local Stability Priors | A coupled computational approach to simultaneously learn a vector field and
the region of attraction of an equilibrium point from generated trajectories of
the system is proposed. The nonlinear identification leverages the local
stability information as a prior on the system, effectively endowing the
estimate with this important structural property. In addition, the knowledge of
the region of attraction plays an experiment design role by informing the
selection of initial conditions from which trajectories are generated and by
enabling the use of a Lyapunov function of the system as a regularization term.
Numerical results show that the proposed method allows efficient sampling and
provides an accurate estimate of the dynamics in an inner approximation of its
region of attraction.
| stat |
Change Detection in Graph Streams by Learning Graph Embeddings on
Constant-Curvature Manifolds | The space of graphs is often characterised by a non-trivial geometry, which
complicates learning and inference in practical applications. A common approach
is to use embedding techniques to represent graphs as points in a conventional
Euclidean space, but non-Euclidean spaces have often been shown to be better
suited for embedding graphs. Among these, constant-curvature Riemannian
manifolds (CCMs) offer embedding spaces suitable for studying the statistical
properties of a graph distribution, as they provide ways to easily compute
metric geodesic distances. In this paper, we focus on the problem of detecting
changes in stationarity in a stream of attributed graphs. To this end, we
introduce a novel change detection framework based on neural networks and CCMs,
that takes into account the non-Euclidean nature of graphs. Our contribution in
this work is twofold. First, via a novel approach based on adversarial
learning, we compute graph embeddings by training an autoencoder to represent
graphs on CCMs. Second, we introduce two novel change detection tests operating
on CCMs. We perform experiments on synthetic data, as well as two real-world
application scenarios: the detection of epileptic seizures using functional
connectivity brain networks, and the detection of hostility between two
subjects, using human skeletal graphs. Results show that the proposed methods
are able to detect even small changes in a graph-generating process,
consistently outperforming approaches based on Euclidean embeddings.
| stat |
A Prototype Knockoff Filter for Group Selection with FDR Control | In many applications, we need to study a linear regression model that
consists of a response variable and a large number of potential explanatory
variables and determine which variables are truly associated with the response.
In 2015, Barber and Candes introduced a new variable selection procedure called
the knockoff filter to control the false discovery rate (FDR) and proved that
this method achieves exact FDR control. In this paper, we propose a prototype
knockoff filter for group selection by extending the Reid-Tibshirani prototype
method. Our prototype knockoff filter improves the computational efficiency and
statistical power of the Reid-Tibshirani prototype method when it is applied
for group selection. In some cases when the group features are spanned by one
or a few hidden factors, we demonstrate that the PCA prototype knockoff filter
outperforms the Dai-Barber group knockoff filter. We present several numerical
experiments to compare our prototype knockoff filter with the Reid-Tibshirani
prototype method and the group knockoff filter. We have also conducted some
analysis of the knockoff filter. Our analysis reveals that some knockoff path
method statistics, including the Lasso path statistic, may lead to loss of
power for certain design matrices and a specially designed response even if
their signal strengths are still relatively strong.
| stat |
How Many Customers Does a Retail Store Have? | The knowledge of the number of customers is the pillar of retail business
analytics. In our setting, we assume that a portion of customers is monitored
and easily counted due to the loyalty program while the rest is not monitored.
The behavior of customers in both groups may significantly differ making the
estimation of the number of unmonitored customers a non-trivial task. We
identify shopping patterns of several customer segments which allows us to
estimate the distribution of customers without the loyalty card using the
maximum likelihood method. In a simulation study, we find that the proposed
approach is quite precise even when the data sample is very small and its
assumptions are violated to a certain degree. In an empirical study of a
drugstore chain, we validate and illustrate the proposed approach in practice.
The actual number of customers estimated by the proposed method is much higher
than the number suggested by the naive estimate assuming the constant customer
distribution. The proposed method can also be utilized to determine penetration
of the loyalty program in the individual customer segments.
| stat |
The network structure of cultural distances | Making use of the information from the World Value Survey (Wave 6), and
operationalizing a definition of national culture that encompasses both the
relevance of specific cultural traits and the interdependence among them, this
paper proposes a methodology to reveal the latent structure of a national
culture and to measure a cultural distance between countries that takes into
account the network structure of national cultural traits. Exploiting the
possibilities offered by Copula graphical models for ordinal and categorical
data, this paper infers the structure of the cultural network of 54 countries
and proposes a new summary measure of national cultural distances. The new
network index of cultural distance shows that, compared to the methodology used
by Inglehart and Welzel (2005), the world appears to be more culturally
heterogeneous than what it was previously measured.
| stat |
Stagewise Enlargement of Batch Size for SGD-based Learning | Existing research shows that the batch size can seriously affect the
performance of stochastic gradient descent~(SGD) based learning, including
training speed and generalization ability. A larger batch size typically
results in less parameter updates. In distributed training, a larger batch size
also results in less frequent communication. However, a larger batch size can
make a generalization gap more easily. Hence, how to set a proper batch size
for SGD has recently attracted much attention. Although some methods about
setting batch size have been proposed, the batch size problem has still not
been well solved. In this paper, we first provide theory to show that a proper
batch size is related to the gap between initialization and optimum of the
model parameter. Then based on this theory, we propose a novel method, called
\underline{s}tagewise \underline{e}nlargement of \underline{b}atch
\underline{s}ize~(\mbox{SEBS}), to set proper batch size for SGD. More
specifically, \mbox{SEBS} adopts a multi-stage scheme, and enlarges the batch
size geometrically by stage. We theoretically prove that, compared to classical
stagewise SGD which decreases learning rate by stage, \mbox{SEBS} can reduce
the number of parameter updates without increasing generalization error. SEBS
is suitable for \mbox{SGD}, momentum \mbox{SGD} and AdaGrad. Empirical results
on real data successfully verify the theories of \mbox{SEBS}. Furthermore,
empirical results also show that SEBS can outperform other baselines.
| stat |
Topological Information Data Analysis | This paper presents methods that quantify the structure of statistical
interactions within a given data set, and was first used in \cite{Tapia2018}.
It establishes new results on the k-multivariate mutual-informations (I_k)
inspired by the topological formulation of Information introduced in. In
particular we show that the vanishing of all I_k for 2\leq k \leq n of n random
variables is equivalent to their statistical independence. Pursuing the work of
Hu Kuo Ting and Te Sun Han, we show that information functions provide
co-ordinates for binary variables, and that they are analytically independent
on the probability simplex for any set of finite variables. The maximal
positive I_k identifies the variables that co-vary the most in the population,
whereas the minimal negative I_k identifies synergistic clusters and the
variables that differentiate-segregate the most the population. Finite data
size effects and estimation biases severely constrain the effective computation
of the information topology on data, and we provide simple statistical tests
for the undersampling bias and the k-dependences following. We give an example
of application of these methods to genetic expression and unsupervised
cell-type classification. The methods unravel biologically relevant subtypes,
with a sample size of 41 genes and with few errors. It establishes generic
basic methods to quantify the epigenetic information storage and a unified
epigenetic unsupervised learning formalism. We propose that higher-order
statistical interactions and non identically distributed variables are
constitutive characteristics of biological systems that should be estimated in
order to unravel their significant statistical structure and diversity. The
topological information data analysis presented here allows to precisely
estimate this higher-order structure characteristic of biological systems.
| stat |
svReg: Structural Varying-coefficient regression to differentiate how
regional brain atrophy affects motor impairment for Huntington disease
severity groups | For Huntington disease, identification of brain regions related to motor
impairment can be useful for developing interventions to alleviate the motor
symptom, the major symptom of the disease. However, the effects from the brain
regions to motor impairment may vary for different groups of patients. Hence,
our interest is not only to identify the brain regions but also to understand
how their effects on motor impairment differ by patient groups. This can be
cast as a model selection problem for a varying-coefficient regression.
However, this is challenging when there is a pre-specified group structure
among variables. We propose a novel variable selection method for a
varying-coefficient regression with such structured variables. Our method is
empirically shown to select relevant variables consistently. Also, our method
screens irrelevant variables better than existing methods. Hence, our method
leads to a model with higher sensitivity, lower false discovery rate and higher
prediction accuracy than the existing methods. Finally, we found that the
effects from the brain regions to motor impairment differ by disease severity
of the patients. To the best of our knowledge, our study is the first to
identify such interaction effects between the disease severity and brain
regions, which indicates the need for customized intervention by disease
severity.
| stat |
Log-Regularly Varying Scale Mixture of Normals for Robust Regression | Linear regression with the classical normality assumption for the error
distribution may lead to an undesirable posterior inference of regression
coefficients due to the potential outliers. This paper considers the finite
mixture of two components with thin and heavy tails as the error distribution,
which has been routinely employed in applied statistics. For the heavily-tailed
component, we introduce the novel class of distributions; their densities are
log-regularly varying and have heavier tails than those of Cauchy distribution,
yet they are expressed as a scale mixture of normal distributions and enable
the efficient posterior inference by Gibbs sampler. We prove the robustness to
outliers of the posterior distributions under the proposed models with a
minimal set of assumptions, which justifies the use of shrinkage priors with
unbounded densities for the coefficient vector in the presence of outliers. The
extensive comparison with the existing methods via simulation study shows the
improved performance of our model in point and interval estimation, as well as
its computational efficiency. Further, we confirm the posterior robustness of
our method in the empirical study with the shrinkage priors for regression
coefficients.
| stat |
Multiple Change Point Detection and Validation in Autoregressive Time
Series Data | It is quite common that the structure of a time series changes abruptly.
Identifying these change points and describing the model structure in the
segments between these change points is of interest. In this paper, time series
data is modelled assuming each segment is an autoregressive time series with
possibly different autoregressive parameters. This is achieved using two main
steps. The first step is to use a likelihood ratio scan based estimation
technique to identify these potential change points to segment the time series.
Once these potential change points are identified, modified parametric spectral
discrimination tests are used to validate the proposed segments. A numerical
study is conducted to demonstrate the performance of the proposed method across
various scenarios and compared against other contemporary techniques.
| stat |
Graph matching beyond perfectly-overlapping Erd\H{o}s--R\'enyi random
graphs | Graph matching is a fruitful area in terms of both algorithms and theories.
In this paper, we exploit the degree information, which was previously used
only in noiseless graphs and perfectly-overlapping Erd\H{o}s--R\'enyi random
graphs matching. We are concerned with graph matching of partially-overlapping
graphs and stochastic block models, which are more useful in tackling real-life
problems. We propose the edge exploited degree profile graph matching method
and two refined varations. We conduct a thorough analysis of our proposed
methods' performances in a range of challenging scenarios, including a
zebrafish neuron activity data set and a coauthorship data set. Our methods are
proved to be numerically superior than the state-of-the-art methods.
| stat |
Understanding Recurrent Neural Networks Using Nonequilibrium Response
Theory | Recurrent neural networks (RNNs) are brain-inspired models widely used in
machine learning for analyzing sequential data. The present work is a
contribution towards a deeper understanding of how RNNs process input signals
using the response theory from nonequilibrium statistical mechanics. For a
class of continuous-time stochastic RNNs (SRNNs) driven by an input signal, we
derive a Volterra type series representation for their output. This
representation is interpretable and disentangles the input signal from the SRNN
architecture. The kernels of the series are certain recursively defined
correlation functions with respect to the unperturbed dynamics that completely
determine the output. Exploiting connections of this representation and its
implications to rough paths theory, we identify a universal feature -- the
response feature, which turns out to be the signature of tensor product of the
input signal and a natural support basis. In particular, we show that SRNNs,
with only the weights in the readout layer optimized and the weights in the
hidden layer kept fixed and not optimized, can be viewed as kernel machines
operating on a reproducing kernel Hilbert space associated with the response
feature.
| stat |
Computationally Efficient Simulation of Queues: The R Package
queuecomputer | Large networks of queueing systems model important real-world systems such as
MapReduce clusters, web-servers, hospitals, call centers and airport passenger
terminals. To model such systems accurately, we must infer queueing parameters
from data. Unfortunately, for many queueing networks there is no clear way to
proceed with parameter inference from data. Approximate Bayesian computation
could offer a straightforward way to infer parameters for such networks if we
could simulate data quickly enough.
We present a computationally efficient method for simulating from a very
general set of queueing networks with the R package queuecomputer. Remarkable
speedups of more than 2 orders of magnitude are observed relative to the
popular DES packages simmer and simpy. We replicate output from these packages
to validate the package.
The package is modular and integrates well with the popular R package dplyr.
Complex queueing networks with tandem, parallel and fork/join topologies can
easily be built with these two packages together. We show how to use this
package with two examples: a call center and an airport terminal.
| stat |
Understanding Approximate Fisher Information for Fast Convergence of
Natural Gradient Descent in Wide Neural Networks | Natural Gradient Descent (NGD) helps to accelerate the convergence of
gradient descent dynamics, but it requires approximations in large-scale deep
neural networks because of its high computational cost. Empirical studies have
confirmed that some NGD methods with approximate Fisher information converge
sufficiently fast in practice. Nevertheless, it remains unclear from the
theoretical perspective why and under what conditions such heuristic
approximations work well. In this work, we reveal that, under specific
conditions, NGD with approximate Fisher information achieves the same fast
convergence to global minima as exact NGD. We consider deep neural networks in
the infinite-width limit, and analyze the asymptotic training dynamics of NGD
in function space via the neural tangent kernel. In the function space, the
training dynamics with the approximate Fisher information are identical to
those with the exact Fisher information, and they converge quickly. The fast
convergence holds in layer-wise approximations; for instance, in block diagonal
approximation where each block corresponds to a layer as well as in block
tri-diagonal and K-FAC approximations. We also find that a unit-wise
approximation achieves the same fast convergence under some assumptions. All of
these different approximations have an isotropic gradient in the function
space, and this plays a fundamental role in achieving the same convergence
properties in training. Thus, the current study gives a novel and unified
theoretical foundation with which to understand NGD methods in deep learning.
| stat |
A Fully Online Approach for Covariance Matrices Estimation of Stochastic
Gradient Descent Solutions | Stochastic gradient descent (SGD) algorithm is widely used for parameter
estimation especially in online setting. While this recursive algorithm is
popular for computation and memory efficiency, the problem of quantifying
variability and randomness of the solutions has been rarely studied. This paper
aims at conducting statistical inference of SGD-based estimates in online
setting. In particular, we propose a fully online estimator for the covariance
matrix of averaged SGD iterates (ASGD). Based on the classic asymptotic
normality results of ASGD, we construct asymptotically valid confidence
intervals for model parameters. Upon receiving new observations, we can quickly
update the covariance estimator and confidence intervals. This approach fits in
online setting even if the total number of data is unknown and takes the full
advantage of SGD: efficiency in both computation and memory.
| stat |
A Simple Descriptive Method & Standard for Comparing Pairs of Stacked
Bar Graphs | While a plethora of research has been devoted to extoling the power and
importance of data visualization, research on the effectiveness of data
visualization methods from a human perceptual, and more generally, a cognitive
standpoint remains largely untapped. Indeed, the way that human observers
perceive and judge graphic charts can determine the interpretation of the
graphed data. In this brief note we introduce a simple method for comparing
stacked bar graphs based on a well-known result from cognitive science.
| stat |
To Ensemble or Not Ensemble: When does End-To-End Training Fail? | End-to-End training (E2E) is becoming more and more popular to train complex
Deep Network architectures. An interesting question is whether this trend will
continue-are there any clear failure cases for E2E training? We study this
question in depth, for the specific case of E2E training an ensemble of
networks. Our strategy is to blend the gradient smoothly in between two
extremes: from independent training of the networks, up to to full E2E
training. We find clear failure cases, where over-parameterized models cannot
be trained E2E. A surprising result is that the optimum can sometimes lie in
between the two, neither an ensemble or an E2E system. The work also uncovers
links to Dropout, and raises questions around the nature of ensemble diversity
and multi-branch networks.
| stat |
Novel and Efficient Approximations for Zero-One Loss of Linear
Classifiers | The predictive quality of machine learning models is typically measured in
terms of their (approximate) expected prediction accuracy or the so-called Area
Under the Curve (AUC). Minimizing the reciprocals of these measures are the
goals of supervised learning. However, when the models are constructed by the
means of empirical risk minimization (ERM), surrogate functions such as the
logistic loss or hinge loss are optimized instead. In this work, we show that
in the case of linear predictors, the expected error and the expected ranking
loss can be effectively approximated by smooth functions whose closed form
expressions and those of their first (and second) order derivatives depend on
the first and second moments of the data distribution, which can be
precomputed. Hence, the complexity of an optimization algorithm applied to
these functions does not depend on the size of the training data. These
approximation functions are derived under the assumption that the output of the
linear classifier for a given data set has an approximately normal
distribution. We argue that this assumption is significantly weaker than the
Gaussian assumption on the data itself and we support this claim by
demonstrating that our new approximation is quite accurate on data sets that
are not necessarily Gaussian. We present computational results that show that
our proposed approximations and related optimization algorithms can produce
linear classifiers with similar or better test accuracy or AUC, than those
obtained using state-of-the-art approaches, in a fraction of the time.
| stat |
Understanding and predicting travel time with spatio-temporal features
of network traffic flow, weather and incidents | Travel time on a route varies substantially by time of day and from day to
day. It is critical to understand to what extent this variation is correlated
with various factors, such as weather, incidents, events or travel demand level
in the context of dynamic networks. This helps a better decision making for
infrastructure planning and real-time traffic operation. We propose a
data-driven approach to understand and predict highway travel time using
spatio-temporal features of those factors, all of which are acquired from
multiple data sources. The prediction model holistically selects the most
related features from a high-dimensional feature space by correlation analysis,
principle component analysis and LASSO. We test and compare the performance of
several regression models in predicting travel time 30 min in advance via two
case studies: (1) a 6-mile highway corridor of I-270N in D.C. region, and (2) a
2.3-mile corridor of I-376E in Pittsburgh region. We found that some
bottlenecks scattered in the network can imply congestion on those corridors at
least 30 minutes in advance, including those on the alternative route to the
corridors of study. In addition, real-time travel time is statistically related
to incidents on some specific locations, morning/afternoon travel demand,
visibility, precipitation, wind speed/gust and the weather type. All those
spatio-temporal information together help improve prediction accuracy,
comparing to using only speed data. In both case studies, random forest shows
the most promise, reaching a root-mean-squared error of 16.6\% and 17.0\%
respectively in afternoon peak hours for the entire year of 2014.
| stat |
Understanding Priors in Bayesian Neural Networks at the Unit Level | We investigate deep Bayesian neural networks with Gaussian weight priors and
a class of ReLU-like nonlinearities. Bayesian neural networks with Gaussian
priors are well known to induce an L2, "weight decay", regularization. Our
results characterize a more intricate regularization effect at the level of the
unit activations. Our main result establishes that the induced prior
distribution on the units before and after activation becomes increasingly
heavy-tailed with the depth of the layer. We show that first layer units are
Gaussian, second layer units are sub-exponential, and units in deeper layers
are characterized by sub-Weibull distributions. Our results provide new
theoretical insight on deep Bayesian neural networks, which we corroborate with
simulation experiments.
| stat |
Distance-learning For Approximate Bayesian Computation To Model a
Volcanic Eruption | Approximate Bayesian computation (ABC) provides us with a way to infer
parameters of models, for which the likelihood function is not available, from
an observation. Using ABC, which depends on many simulations from the
considered model, we develop an inferential framework to learn parameters of a
stochastic numerical simulator of volcanic eruption. Moreover, the model itself
is parallelized using Message Passing Interface (MPI). Thus, we develop a
nested-parallelized MPI communicator to handle the expensive numerical model
with ABC algorithms. ABC usually relies on summary statistics of the data in
order to measure the discrepancy model output and observation. However,
informative summary statistics cannot be found for the considered model. We
therefore develop a technique to learn a distance between model outputs based
on deep metric-learning. We use this framework to learn the plume
characteristics (eg. initial plume velocity) of the volcanic eruption from the
tephra deposits collected by field-work associated with the 2450 BP Pululagua
(Ecuador) volcanic eruption.
| stat |
Spike-and-Slab Meets LASSO: A Review of the Spike-and-Slab LASSO | High-dimensional data sets have become ubiquitous in the past few decades,
often with many more covariates than observations. In the frequentist setting,
penalized likelihood methods are the most popular approach for variable
selection and estimation in high-dimensional data. In the Bayesian framework,
spike-and-slab methods are commonly used as probabilistic constructs for
high-dimensional modeling. Within the context of linear regression, Rockova and
George (2018) introduced the spike-and-slab LASSO (SSL), an approach based on a
prior which provides a continuum between the penalized likelihood LASSO and the
Bayesian point-mass spike-and-slab formulations. Since its inception, the
spike-and-slab LASSO has been extended to a variety of contexts, including
generalized linear models, factor analysis, graphical models, and nonparametric
regression. The goal of this paper is to survey the landscape surrounding
spike-and-slab LASSO methodology. First we elucidate the attractive properties
and the computational tractability of SSL priors in high dimensions. We then
review methodological developments of the SSL and outline several theoretical
developments. We illustrate the methodology on both simulated and real
datasets.
| stat |
Modeling Risk via Realized HYGARCH Model | In this paper, we propose the realized Hyperbolic GARCH model for the
joint-dynamics of lowfrequency returns and realized measures that generalizes
the realized GARCH model of Hansen et al.(2012) as well as the FLoGARCH model
introduced by Vander Elst (2015). This model is sufficiently flexible to
capture both long memory and asymmetries related to leverage effects. In
addition, we will study the strictly and weak stationarity conditions of the
model. To evaluate its performance, experimental simulations, using the Monte
Carlo method, are made to forecast the Value at Risk (VaR) and the Expected
Shortfall (ES). These simulation studies show that for ES and VaR forecasting,
the realized Hyperbolic GARCH (RHYGARCH-GG) model with Gaussian-Gaussian errors
provide more adequate estimates than the realized Hyperbolic GARCH model with
student- Gaussian errors.
| stat |
Simulating Tariff Impact in Electrical Energy Consumption Profiles with
Conditional Variational Autoencoders | The implementation of efficient demand response (DR) programs for household
electricity consumption would benefit from data-driven methods capable of
simulating the impact of different tariffs schemes. This paper proposes a novel
method based on conditional variational autoencoders (CVAE) to generate, from
an electricity tariff profile combined with exogenous weather and calendar
variables, daily consumption profiles of consumers segmented in different
clusters. First, a large set of consumers is gathered into clusters according
to their consumption behavior and price-responsiveness. The clustering method
is based on a causality model that measures the effect of a specific tariff on
the consumption level. Then, daily electrical energy consumption profiles are
generated for each cluster with CVAE. This non-parametric approach is compared
to a semi-parametric data generator based on generalized additive models and
that uses prior knowledge of energy consumption. Experiments in a publicly
available data set show that, the proposed method presents comparable
performance to the semi-parametric one when it comes to generating the average
value of the original data. The main contribution from this new method is the
capacity to reproduce rebound and side effects in the generated consumption
profiles. Indeed, the application of a special electricity tariff over a time
window may also affect consumption outside this time window. Another
contribution is that the clustering approach segments consumers according to
their daily consumption profile and elasticity to tariff changes. These two
results combined are very relevant for an ex-ante testing of future DR policies
by system operators, retailers and energy regulators.
| stat |
Bayesian Survival Analysis Using Gamma Processes with Adaptive Time
Partition | In Bayesian semi-parametric analyses of time-to-event data, non-parametric
process priors are adopted for the baseline hazard function or the cumulative
baseline hazard function for a given finite partition of the time axis.
However, it would be controversial to suggest a general guideline to construct
an optimal time partition. While a great deal of research has been done to
relax the assumption of the fixed split times for other non-parametric
processes, to our knowledge, no methods have been developed for a gamma process
prior, which is one of the most widely used in Bayesian survival analysis. In
this paper, we propose a new Bayesian framework for proportional hazards models
where the cumulative baseline hazard function is modeled a priori by a gamma
process. A key feature of the proposed framework is that the number and
position of interval cutpoints are treated as random and estimated based on
their posterior distributions.
| stat |
Covariate Shift Adaptation in High-Dimensional and Divergent
Distributions | In real world applications of supervised learning methods, training and test
sets are often sampled from the distinct distributions and we must resort to
domain adaptation techniques. One special class of techniques is Covariate
Shift Adaptation, which allows practitioners to obtain good generalization
performance in the distribution of interest when domains differ only by the
marginal distribution of features. Traditionally, Covariate Shift Adaptation is
implemented using Importance Weighting which may fail in high-dimensional
settings due to small Effective Sample Sizes (ESS). In this paper, we propose
(i) a connection between ESS, high-dimensional settings and generalization
bounds and (ii) a simple, general and theoretically sound approach to combine
feature selection and Covariate Shift Adaptation. The new approach yields good
performance with improved ESS.
| stat |
MIMIC modelling with instrumental variables: A 2SLS-MIMIC approach | Multiple Indicators Multiple Causes (MIMIC) models are type of structural
equation models, a theory-based approach to confirm the influence of a set of
exogenous causal variables on the latent variable, and also the effect of the
latent variable on observed indicator variables. In a common MIMIC model,
multiple indicators reflect the underlying latent variables/factors, and the
multiple causes (observed predictors) affect latent variables/factors. Basic
assumptions of MIMIC are clearly violated in case of a variable being both an
indicator and a cause, i.e. in the presence of reverse causality. Furthermore,
the model is then unidentified. To resolve the situation, which can arise
frequently, and as MIMIC estimation lacks closed form solutions for parameters
we utilize a version of Bollen's (1996) 2SLS estimator for structural equation
models combined with J\"oreskog (1970)'s method of the analysis of covariance
structures to derive a new, 2SLS estimator for MIMIC models. Our 2SLS empirical
estimation is based on static MIMIC specification but we point also to
dynamic/error-correction MIMIC specification and 2SLS solution for it. We
derive basic asymptotic theory for static 2SLS-MIMIC, present a simulation
study and apply findings to an interesting empirical case of estimating
precarious status of older workers (using dataset of Survey of Health, Ageing
and Retirement in Europe) which solves an important issue of the definition of
precarious work as a multidimensional concept, not modelled adequately so far.
| stat |
Binary Outcome Copula Regression Model with Sampling Gradient Fitting | Use copula to model dependency of variable extends multivariate gaussian
assumption. In this paper we first empirically studied copula regression model
with continous response. Both simulation study and real data study are given.
Secondly we give a novel copula regression model with binary outcome, and we
propose a score gradient estimation algorithms to fit the model. Both
simulation study and real data study are given for our model and fitting
algorithm.
| stat |
Fourier Phase Retrieval with Extended Support Estimation via Deep Neural
Network | We consider the problem of sparse phase retrieval from Fourier transform
magnitudes to recover the $k$-sparse signal vector and its support
$\mathcal{T}$. We exploit extended support estimate $\mathcal{E}$ with size
larger than $k$ satisfying $\mathcal{E} \supseteq \mathcal{T}$ and obtained by
a trained deep neural network (DNN). To make the DNN learnable, it provides
$\mathcal{E}$ as the union of equivalent solutions of $\mathcal{T}$ by
utilizing modulo Fourier invariances. Set $\mathcal{E}$ can be estimated with
short running time via the DNN, and support $\mathcal{T}$ can be determined
from the DNN output rather than from the full index set by applying hard
thresholding to $\mathcal{E}$. Thus, the DNN-based extended support estimation
improves the reconstruction performance of the signal with a low complexity
burden dependent on $k$. Numerical results verify that the proposed scheme has
a superior performance with lower complexity compared to local search-based
greedy sparse phase retrieval and a state-of-the-art variant of the Fienup
method.
| stat |
gfpop: an R Package for Univariate Graph-Constrained Change-point
Detection | In a world with data that change rapidly and abruptly, it is important to
detect those changes accurately. In this paper we describe an R package
implementing an algorithm recently proposed by Hocking et al. [2017] for
penalised maximum likelihood inference of constrained multiple change-point
models. This algorithm can be used to pinpoint the precise locations of abrupt
changes in large data sequences. There are many application domains for such
models, such as medicine, neuroscience or genomics. Often, practitioners have
prior knowledge about the changes they are looking for. For example in genomic
data, biologists sometimes expect peaks: up changes followed by down changes.
Taking advantage of such prior information can substantially improve the
accuracy with which we can detect and estimate changes. Hocking et al. [2017]
described a graph framework to encode many examples of such prior information
and a generic algorithm to infer the optimal model parameters, but implemented
the algorithm for just a single scenario. We present the gfpop package that
implements the algorithm in a generic manner in R/C++. gfpop works for a
user-defined graph that can encode the prior nformation of the types of change
and implements several loss functions (Gauss, Poisson, Binomial, Biweight and
Huber). We then illustrate the use of gfpop on isotonic simulations and several
applications in biology. For a number of graphs the algorithm runs in a matter
of seconds or minutes for 10^5 datapoints.
| stat |
Modeling Oral Glucose Tolerance Test (OGTT) data and its Bayesian
Inverse Problem | One common way to test for diabetes is the Oral Glucose Tolerance Test or
OGTT. Most common methods for the analysis of the data on this test are
wasteful of much of the information contained therein. We propose to model
blood glucose during an OGTT using a compartmental dynamic model with a system
of ODEs. Our model works well in describing most scenarios that occur during an
OGTT considering only 4 parameters. Fitting the model to data is an inverse
problem, which is suitable for Bayesian inference. Priors are specified and
posterior inference results are shown using real data.
| stat |
Deep Two-Way Matrix Reordering for Relational Data Analysis | Matrix reordering is a task to permute the rows and columns of a given
observed matrix such that the resulting reordered matrix shows meaningful or
interpretable structural patterns. Most existing matrix reordering techniques
share the common processes of extracting some feature representations from an
observed matrix in a predefined manner, and applying matrix reordering based on
it. However, in some practical cases, we do not always have prior knowledge
about the structural pattern of an observed matrix. To address this problem, we
propose a new matrix reordering method, called deep two-way matrix reordering
(DeepTMR), using a neural network model. The trained network can automatically
extract nonlinear row/column features from an observed matrix, which can then
be used for matrix reordering. Moreover, the proposed DeepTMR provides the
denoised mean matrix of a given observed matrix as an output of the trained
network. This denoised mean matrix can be used to visualize the global
structure of the reordered observed matrix. We demonstrate the effectiveness of
the proposed DeepTMR by applying it to both synthetic and practical datasets.
| stat |
Policy Implications of Statistical Estimates: A General Bayesian
Decision-Theoretic Model for Binary Outcomes | How should scholars evaluate the statistically estimated causal effect of a
policy intervention? I point out three limitations in the conventional
practice. First, relying on statistical significance misses the fact that
uncertainty is a continuous scale. Second, focusing on a standard point
estimate overlooks variation in plausible effect sizes. Third, the criterion of
substantive significance is rarely explained or justified. To address these
issues, I propose an original Bayesian decision-theoretic model for binary
outcomes. I incorporate the posterior distribution of a causal effect reducing
the likelihood of an undesirable event, into a loss function over the cost of a
policy to realize the effect and the cost of the event. The model can use an
effect size of interest other than the standard point estimate, and the
probability of this effect as a continuous measure of uncertainty. It then
presents approximately up to what ratio between the two costs an expected loss
remains smaller if the policy is implemented than if not. I exemplify my model
through three applications and provide an R package for easy implementation.
| stat |
A probabilistic gridded product for daily precipitation extremes over
the United States | Gridded data products, for example interpolated daily measurements of
precipitation from weather stations, are commonly used as a convenient
substitute for direct observations because these products provide a spatially
and temporally continuous and complete source of data. However, when the goal
is to characterize climatological features of extreme precipitation over a
spatial domain (e.g., a map of return values) at the native spatial scales of
these phenomena, then gridded products may lead to incorrect conclusions
because daily precipitation is a fractal field and hence any smoothing
technique will dampen local extremes. To address this issue, we create a new
"probabilistic" gridded product specifically designed to characterize the
climatological properties of extreme precipitation by applying spatial
statistical analyses to daily measurements of precipitation from the GHCN over
CONUS. The essence of our method is to first estimate the climatology of
extreme precipitation based on station data and then use a data-driven
statistical approach to interpolate these estimates to a fine grid. We argue
that our method yields an improved characterization of the climatology within a
grid cell because the probabilistic behavior of extreme precipitation is much
better behaved (i.e., smoother) than daily weather. Furthermore, the spatial
smoothing innate to our approach significantly increases the signal-to-noise
ratio in the estimated extreme statistics relative to an analysis without
smoothing. Finally, by deriving a data-driven approach for translating extreme
statistics to a spatially complete grid, the methodology outlined in this paper
resolves the issue of how to properly compare station data with output from
earth system models. We conclude the paper by comparing our probabilistic
gridded product with a standard extreme value analysis of the Livneh gridded
daily precipitation product.
| stat |
Optimal Subsampling Algorithms for Big Data Regressions | To fast approximate maximum likelihood estimators with massive data, this
paper studies the Optimal Subsampling Method under the A-optimality Criterion
(OSMAC) for generalized linear models. The consistency and asymptotic normality
of the estimator from a general subsampling algorithm are established, and
optimal subsampling probabilities under the A- and L-optimality criteria are
derived. Furthermore, using Frobenius norm matrix concentration inequalities,
finite sample properties of the subsample estimator based on optimal
subsampling probabilities are also derived. Since the optimal subsampling
probabilities depend on the full data estimate, an adaptive two-step algorithm
is developed. Asymptotic normality and optimality of the estimator from this
adaptive algorithm are established. The proposed methods are illustrated and
evaluated through numerical experiments on simulated and real datasets.
| stat |
Accelerating Convergence of Replica Exchange Stochastic Gradient MCMC
via Variance Reduction | Replica exchange stochastic gradient Langevin dynamics (reSGLD) has shown
promise in accelerating the convergence in non-convex learning; however, an
excessively large correction for avoiding biases from noisy energy estimators
has limited the potential of the acceleration. To address this issue, we study
the variance reduction for noisy energy estimators, which promotes much more
effective swaps. Theoretically, we provide a non-asymptotic analysis on the
exponential acceleration for the underlying continuous-time Markov jump
process; moreover, we consider a generalized Girsanov theorem which includes
the change of Poisson measure to overcome the crude discretization based on the
Gr\"{o}wall's inequality and yields a much tighter error in the 2-Wasserstein
($\mathcal{W}_2$) distance. Numerically, we conduct extensive experiments and
obtain the state-of-the-art results in optimization and uncertainty estimates
for synthetic experiments and image data.
| stat |
Scrutinizing and De-Biasing Intuitive Physics with Neural Stethoscopes | Visually predicting the stability of block towers is a popular task in the
domain of intuitive physics. While previous work focusses on prediction
accuracy, a one-dimensional performance measure, we provide a broader analysis
of the learned physical understanding of the final model and how the learning
process can be guided. To this end, we introduce neural stethoscopes as a
general purpose framework for quantifying the degree of importance of specific
factors of influence in deep neural networks as well as for actively promoting
and suppressing information as appropriate. In doing so, we unify concepts from
multitask learning as well as training with auxiliary and adversarial losses.
We apply neural stethoscopes to analyse the state-of-the-art neural network for
stability prediction. We show that the baseline model is susceptible to being
misled by incorrect visual cues. This leads to a performance breakdown to the
level of random guessing when training on scenarios where visual cues are
inversely correlated with stability. Using stethoscopes to promote meaningful
feature extraction increases performance from 51% to 90% prediction accuracy.
Conversely, training on an easy dataset where visual cues are positively
correlated with stability, the baseline model learns a bias leading to poor
performance on a harder dataset. Using an adversarial stethoscope, the network
is successfully de-biased, leading to a performance increase from 66% to 88%.
| stat |
On Computationally-Scalable Spatio-Temporal Regression Clustering of
Precipitation Threshold Excesses | Focusing on regression based analysis of extremes in a presence of
systematically missing covariates, this work presents a data-driven
spatio-temporal regression based clustering of threshold excesses. It is shown
that in a presence of systematically missing covariates the behavior of
threshold excesses becomes nonstationary and nonhomogenous. The presented
approach describes this complex behavior by a set of local stationary
Generalized Pareto Distribution (GPD) models, where the parameters are
expressed as regression models, and a latent spatio-temporal switching process.
The spatio-temporal switching process is resolved by the nonparametric Finite
Element Methodology for time series analysis with Bounded Variation of the
model parameters (FEM-BV). The presented FEM-BV-GPD approach goes beyond strong
a priori assumptions made in standard latent class models like Mixture Models
and Hidden Markov Models. In addition, it provides a pragmatic description of
the underlying dependency structure. The performance of the framework is
demonstrated on historical precipitation data for Switzerland and compared with
the results obtained by the standard methods on the same data.
| stat |
Causal Bayesian Optimization | This paper studies the problem of globally optimizing a variable of interest
that is part of a causal model in which a sequence of interventions can be
performed. This problem arises in biology, operational research, communications
and, more generally, in all fields where the goal is to optimize an output
metric of a system of interconnected nodes. Our approach combines ideas from
causal inference, uncertainty quantification and sequential decision making. In
particular, it generalizes Bayesian optimization, which treats the input
variables of the objective function as independent, to scenarios where causal
information is available. We show how knowing the causal graph significantly
improves the ability to reason about optimal decision making strategies
decreasing the optimization cost while avoiding suboptimal solutions. We
propose a new algorithm called Causal Bayesian Optimization (CBO). CBO
automatically balances two trade-offs: the classical exploration-exploitation
and the new observation-intervention, which emerges when combining real
interventional data with the estimated intervention effects computed via
do-calculus. We demonstrate the practical benefits of this method in a
synthetic setting and in two real-world applications.
| stat |
Faster Uncertainty Quantification for Inverse Problems with Conditional
Normalizing Flows | In inverse problems, we often have access to data consisting of paired
samples $(x,y)\sim p_{X,Y}(x,y)$ where $y$ are partial observations of a
physical system, and $x$ represents the unknowns of the problem. Under these
circumstances, we can employ supervised training to learn a solution $x$ and
its uncertainty from the observations $y$. We refer to this problem as the
"supervised" case. However, the data $y\sim p_{Y}(y)$ collected at one point
could be distributed differently than observations $y'\sim p_{Y}'(y')$,
relevant for a current set of problems. In the context of Bayesian inference,
we propose a two-step scheme, which makes use of normalizing flows and joint
data to train a conditional generator $q_{\theta}(x|y)$ to approximate the
target posterior density $p_{X|Y}(x|y)$. Additionally, this preliminary phase
provides a density function $q_{\theta}(x|y)$, which can be recast as a prior
for the "unsupervised" problem, e.g.~when only the observations $y'\sim
p_{Y}'(y')$, a likelihood model $y'|x$, and a prior on $x'$ are known. We then
train another invertible generator with output density $q'_{\phi}(x|y')$
specifically for $y'$, allowing us to sample from the posterior
$p_{X|Y}'(x|y')$. We present some synthetic results that demonstrate
considerable training speedup when reusing the pretrained network
$q_{\theta}(x|y')$ as a warm start or preconditioning for approximating
$p_{X|Y}'(x|y')$, instead of learning from scratch. This training modality can
be interpreted as an instance of transfer learning. This result is particularly
relevant for large-scale inverse problems that employ expensive numerical
simulations.
| stat |
Variational Bayes for Gaussian Factor Models under the Cumulative
Shrinkage Process | The cumulative shrinkage process is an increasing shrinkage prior that can be
employed within models in which additional terms are supposed to play a
progressively negligible role. A natural application is to Gaussian factor
models, where such a process has proved effective in inducing parsimonious
representations while providing accurate inference on the data covariance
matrix. The cumulative shrinkage process came with an adaptive Gibbs sampler
that tunes the number of latent factors throughout iterations, which makes it
faster than the non-adaptive Gibbs sampler. In this work we propose a
variational algorithm for Gaussian factor models endowed with a cumulative
shrinkage process. Such a strategy provides comparable inference with respect
to the adaptive Gibbs sampler and further reduces runtime
| stat |
Smooth Adjustment for Correlated Effects | This paper considers a high dimensional linear regression model with
corrected variables. A variety of methods have been developed in recent years,
yet it is still challenging to keep accurate estimation when there are complex
correlation structures among predictors and the response. We propose an
adaptive and "reversed" penalty for regularization to solve this problem. This
penalty doesn't shrink variables but focuses on removing the shrinkage bias and
encouraging grouping effect. Combining the l_1 penalty and the Minimax Concave
Penalty (MCP), we propose two methods called Smooth Adjustment for Correlated
Effects (SACE) and Generalized Smooth Adjustment for Correlated Effects
(GSACE). Compared with the traditional adaptive estimator, the proposed methods
have less influence from the initial estimator and can reduce the false
negatives of the initial estimation. The proposed methods can be seen as linear
functions of the new penalty's tuning parameter, and are shown to estimate the
coefficients accurately in both extremely highly correlated variables situation
and weakly correlated variables situation. Under mild regularity conditions we
prove that the methods satisfy certain oracle property. We show by simulations
and applications that the proposed methods often outperforms other methods.
| stat |
Forecasting: theory and practice | Forecasting has always been in the forefront of decision making and planning.
The uncertainty that surrounds the future is both exciting and challenging,
with individuals and organisations seeking to minimise risks and maximise
utilities. The lack of a free-lunch theorem implies the need for a diverse set
of forecasting methods to tackle an array of applications. This unique article
provides a non-systematic review of the theory and the practice of forecasting.
We offer a wide range of theoretical, state-of-the-art models, methods,
principles, and approaches to prepare, produce, organise, and evaluate
forecasts. We then demonstrate how such theoretical concepts are applied in a
variety of real-life contexts, including operations, economics, finance,
energy, environment, and social good. We do not claim that this review is an
exhaustive list of methods and applications. The list was compiled based on the
expertise and interests of the authors. However, we wish that our encyclopedic
presentation will offer a point of reference for the rich work that has been
undertaken over the last decades, with some key insights for the future of the
forecasting theory and practice.
| stat |
Statistical power for cluster analysis | Cluster algorithms are gaining in popularity due to their compelling ability
to identify discrete subgroups in data, and their increasing accessibility in
mainstream programming languages and statistical software. While researchers
can follow guidelines to choose the right algorithms, and to determine what
constitutes convincing clustering, there are no firmly established ways of
computing a priori statistical power for cluster analysis. Here, we take a
simulation approach to estimate power and classification accuracy for popular
analysis pipelines. We systematically varied cluster size, number of clusters,
number of different features between clusters, effect size within each
different feature, and cluster covariance structure in generated datasets. We
then subjected these datasets to common dimensionality reduction approaches
(none, multi-dimensional scaling, or uniform manifold approximation and
projection) and cluster algorithms (k-means, hierarchical agglomerative
clustering with Ward linkage and Euclidean distance, or average linkage and
cosine distance, HDBSCAN). Furthermore, we simulated additional datasets to
explore the effect of sample size and cluster separation on statistical power
and classification accuracy. We found that clustering outcomes were driven by
large effect sizes or the accumulation of many smaller effects across features,
and were mostly unaffected by differences in covariance structure. Sufficient
statistical power can be achieved with relatively small samples (N=20 per
subgroup), provided cluster separation is large ({\Delta}=4). Finally, we
discuss whether fuzzy clustering (c-means) could provide a more parsimonious
alternative for identifying separable multivariate normal distributions,
particularly those with lower centroid separation.
| stat |
Survey Data and Human Computation for Improved Flu Tracking | While digital trace data from sources like search engines hold enormous
potential for tracking and understanding human behavior, these streams of data
lack information about the actual experiences of those individuals generating
the data. Moreover, most current methods ignore or under-utilize human
processing capabilities that allow humans to solve problems not yet solvable by
computers (human computation). We demonstrate how behavioral research, linking
digital and real-world behavior, along with human computation, can be utilized
to improve the performance of studies using digital data streams. This study
looks at the use of search data to track prevalence of Influenza-Like Illness
(ILI). We build a behavioral model of flu search based on survey data linked to
users online browsing data. We then utilize human computation for classifying
search strings. Leveraging these resources, we construct a tracking model of
ILI prevalence that outperforms strong historical benchmarks using only a
limited stream of search data and lends itself to tracking ILI in smaller
geographic units. While this paper only addresses searches related to ILI, the
method we describe has potential for tracking a broad set of phenomena in near
real-time.
| stat |
Bayesian Models for Prediction of the Set-Difference in Volleyball | The aim of this paper is to study and develop Bayesian models for the
analysis of volleyball match outcomes as recorded by the set-difference.
Due to the peculiarity of the outcome variable (set-difference) which takes
discrete values from $-3$ to $3$, we cannot consider standard models based on
the usual Poisson or binomial assumptions used for other sports such as
football/soccer. Hence, the first and foremost challenge was to build models
appropriate for the set-differences of each volleyball match. Here we consider
two major approaches:
a) an ordered multinomial logistic regression model and
b) a model based on a truncated version of the Skellam distribution.
For the first model, we consider the set-difference as an ordinal response
variable within the framework of multinomial logistic regression models.
Concerning the second model, we adjust the Skellam distribution in order to
account for the volleyball rules.
We fit and compare both models with the same covariate structure as in Karlis
& Ntzoufras (2003).
Both models are fitted, illustrated and compared within Bayesian framework
using data from both the regular season and the play-offs of the season 2016/17
of the Greek national men's volleyball league A1.
| stat |
A New Covariance Estimator for Sufficient Dimension Reduction in
High-Dimensional and Undersized Sample Problems | The application of standard sufficient dimension reduction methods for
reducing the dimension space of predictors without losing regression
information requires inverting the covariance matrix of the predictors. This
has posed a number of challenges especially when analyzing high-dimensional
data sets in which the number of predictors $\mathit{p}$ is much larger than
number of samples $n,~(n\ll p)$. A new covariance estimator, called the
\textit{Maximum Entropy Covariance} (MEC) that addresses loss of covariance
information when similar covariance matrices are linearly combined using
\textit{Maximum Entropy} (ME) principle is proposed in this work. By
benefitting naturally from slicing or discretizing range of the response
variable, y into \textit{H} non-overlapping categories, $\mathit{h_{1},\ldots
,h_{H}}$, MEC first combines covariance matrices arising from samples in each y
slice $\mathit{h\in H}$ and then select the one that maximizes entropy under
the principle of maximum uncertainty. The MEC estimator is then formed from
convex mixture of such entropy-maximizing sample covariance $S_{\mbox{mec}}$
estimate and pooled sample covariance $\mathbf{S}_{\mathit{p}}$ estimate across
the $\mathit{H}$ slices without requiring time-consuming covariance
optimization procedures. MEC deals directly with singularity and instability of
sample group covariance estimate in both regression and classification
problems. The efficiency of the MEC estimator is studied with the existing
sufficient dimension reduction methods such as \textit{Sliced Inverse
Regression} (SIR) and \textit{Sliced Average Variance Estimator} (SAVE) as
demonstrated on both classification and regression problems using real life
Leukemia cancer data and customers' electricity load profiles from smart meter
data sets respectively.
| stat |
Non-parametric regression models for compositional data | Compositional data arise in many real-life applications and versatile methods
for properly analyzing this type of data in the regression context are needed.
This paper, through use of the $\alpha$-transformation, extends the classical
$k$-$NN$ regression to what is termed $\alpha$-$k$-$NN$ regression, yielding a
highly flexible non-parametric regression model for compositional data. The
$\alpha$-$k$-$NN$ is further extended to the $\alpha$-kernel regression by
adopting the Nadaray-Watson estimator. Unlike many of the recommended
regression models for compositional data, zeros values (which commonly occur in
practice) are not problematic and they can be incorporated into the proposed
models without modification. Extensive simulation studies and real-life data
analyses highlight the advantage of using these non-parametric regressions for
complex relationships between the compositional response data and Euclidean
predictor variables. Both suggest that $\alpha$-$k$-$NN$ and $\alpha$-kernel
regressions can lead to more accurate predictions compared to current
regression models which assume a, sometimes restrictive, parametric
relationship with the predictor variables. In addition, the $\alpha$-$k$-$NN$
regression, in contrast to $\alpha$-kernel regression, enjoys a high
computational efficiency rendering it highly attractive for use with large
scale, massive, or big data.
| stat |
Bayesian causal inference with some invalid instrumental variables | In observational studies, instrumental variables estimation is greatly
utilized to identify causal effects. One of the key conditions for the
instrumental variables estimator to be consistent is the exclusion restriction,
which indicates that instruments affect the outcome of interest only via the
exposure variable of interest. We propose a likelihood-free Bayesian approach
to make consistent inferences about the causal effect when there are some
invalid instruments in a way that they violate the exclusion restriction
condition. Asymptotic properties of the proposed Bayes estimator, including
consistency and normality, are established. A simulation study demonstrates
that the proposed Bayesian method produces consistent point estimators and
valid credible intervals with correct coverage rates for Gaussian and
non-Gaussian data with some invalid instruments. We also demonstrate the
proposed method through the real data application.
| stat |
A Bayesian Solution to the M-Bias Problem | It is common practice in using regression type models for inferring causal
effects, that inferring the correct causal relationship requires extra
covariates are included or ``adjusted for''. Without performing this adjustment
erroneous causal effects can be inferred. Given this phenomenon it is common
practice to include as many covariates as possible, however such advice comes
unstuck in the presence of M-bias. M-Bias is a problem in causal inference
where the correct estimation of treatment effects requires that certain
variables are not adjusted for i.e. are simply neglected from inclusion in the
model. This issue caused a storm of controversy in 2009 when Rubin, Pearl and
others disagreed about if it could be problematic to include additional
variables in models when inferring causal effects. This paper makes two
contributions to this issue. Firstly we provide a Bayesian solution to the
M-Bias problem. The solution replicates Pearl's solution, but consistent with
Rubin's advice we condition on all variables. Secondly the fact that we are
able to offer a solution to this problem in Bayesian terms shows that it is
indeed possible to represent causal relationships within the Bayesian paradigm,
albeit in an extended space. We make several remarks on the similarities and
differences between causal graphical models which implement the do-calculus and
probabilistic graphical models which enable Bayesian statistics. We hope this
work will stimulate more research on unifying Pearl's causal calculus using
causal graphical models with traditional Bayesian statistics and probabilistic
graphical models.
| stat |
Efficient Social Distancing for COVID-19: An Integration of Economic
Health and Public Health | Social distancing has been the only effective way to contain the spread of an
infectious disease prior to the availability of the pharmaceutical treatment.
It can lower the infection rate of the disease at the economic cost. A pandemic
crisis like COVID-19, however, has posed a dilemma to the policymakers since a
long-term restrictive social distancing or even lockdown will keep economic
cost rising. This paper investigates an efficient social distancing policy to
manage the integrated risk from economic health and public health issues for
COVID-19 using a stochastic epidemic modeling with mobility controls. The
social distancing is to restrict the community mobility, which was recently
accessible with big data analytics. This paper takes advantage of the community
mobility data to model the COVID-19 processes and infer the COVID-19 driven
economic values from major market index price, which allow us to formulate the
search of the efficient social distancing policy as a stochastic control
problem. We propose to solve the problem with a deep-learning approach. By
applying our framework to the US data, we empirically examine the efficiency of
the US social distancing policy and offer recommendations generated from the
algorithm.
| stat |
Training individually fair ML models with Sensitive Subspace Robustness | We consider training machine learning models that are fair in the sense that
their performance is invariant under certain sensitive perturbations to the
inputs. For example, the performance of a resume screening system should be
invariant under changes to the gender and/or ethnicity of the applicant. We
formalize this notion of algorithmic fairness as a variant of individual
fairness and develop a distributionally robust optimization approach to enforce
it during training. We also demonstrate the effectiveness of the approach on
two ML tasks that are susceptible to gender and racial biases.
| stat |
Pervasive Label Errors in Test Sets Destabilize Machine Learning
Benchmarks | We algorithmically identify label errors in the test sets of 10 of the most
commonly-used computer vision, natural language, and audio datasets, and
subsequently study the potential for these label errors to affect benchmark
results. Errors in test sets are numerous and widespread: we estimate an
average of 3.4% errors across the 10 datasets, where for example 2916 label
errors comprise 6% of the ImageNet validation set. Putative label errors are
found using confident learning and then human-validated via crowdsourcing (54%
of the algorithmically-flagged candidates are indeed erroneously labeled).
Surprisingly, we find that lower capacity models may be practically more useful
than higher capacity models in real-world datasets with high proportions of
erroneously labeled data. For example, on ImageNet with corrected labels:
ResNet-18 outperforms ResNet-50 if the prevalence of originally mislabeled test
examples increases by just 6%. On CIFAR-10 with corrected labels: VGG-11
outperforms VGG-19 if the prevalence of originally mislabeled test examples
increases by 5%. Traditionally, ML practitioners choose which model to deploy
based on test accuracy -- our findings advise caution here, proposing that
judging models over correctly labeled test sets may be more useful, especially
for noisy real-world datasets.
| stat |
Fitting Heterogeneous Lanchester Models on the Kursk Campaign | The battle of Kursk between Soviet and German is known to be the biggest tank
battle in the history. The present paper uses the tank and artillery data from
the Kursk database for fitting both forms of homogeneous and heterogeneous
Lanchester model. Under homogeneous form the Soviet (or German) tank casualty
is attributed to only the German(or Soviet) tank engagement. For heterogeneous
form the tank casualty is attributed to both tank and artillery engagements. A
set of differential equations using both forms have been developed, and the
commonly used least square estimation is compared with maximum likelihood
estimation for attrition rates and exponent coefficients. For validating the
models, different goodness-of-fit measures like R2, sum-of-square-residuals
(SSR), root-mean-square error (RMSE), Kolmogorov-Smirnov (KS) and chi-square
statistics are used for comparison. Numerical results suggest the model is
statistically more accurate when each day of the battle is considered as a
mini-battle. The distribution patterns of the SSR and likelihood values with
varying parameters are represented using contour plots and 3D surfaces.
| stat |
Statistical Inference for Networks of High-Dimensional Point Processes | Fueled in part by recent applications in neuroscience, the multivariate
Hawkes process has become a popular tool for modeling the network of
interactions among high-dimensional point process data. While evaluating the
uncertainty of the network estimates is critical in scientific applications,
existing methodological and theoretical work has primarily addressed
estimation. To bridge this gap, this paper develops a new statistical inference
procedure for high-dimensional Hawkes processes. The key ingredient for this
inference procedure is a new concentration inequality on the first- and
second-order statistics for integrated stochastic processes, which summarize
the entire history of the process. Combining recent results on martingale
central limit theory with the new concentration inequality, we then
characterize the convergence rate of the test statistics. We illustrate finite
sample validity of our inferential tools via extensive simulations and
demonstrate their utility by applying them to a neuron spike train data set.
| stat |
Modeling of Covid-19 Pandemic in Cyprus | The Republic of Cyprus is a small island in the southeast of Europe and
member of the European Union. The first wave of COVID-19 in Cyprus started in
early March, 2020 (imported cases) and peaked in late March-early April. The
health authorities responded rapidly and rigorously to the COVID-19 pandemic by
scaling-up testing, increasing efforts to trace and isolate contacts of cases,
and implementing measures such as closures of educational institutions, and
travel and movement restrictions. The pandemic was also a unique opportunity
that brought together experts from various disciplines including
epidemiologists, clinicians, mathematicians, and statisticians. The aim of this
paper is to present the efforts of this new, multidisciplinary research team in
modelling the COVID-19 pandemic in the Republic of Cyprus.
| stat |
Factors associated with injurious from falls in people with early stage
Parkinson's disease | Falls are common in people with Parkinson's disease (PD) and have detrimental
effects which can lower the quality of life. While studies have been conducted
to learn about falling in general, factors distinguishing injurious from
non-injurious falls are less clear. We develop a two-stage Bayesian logistic
regression model was used to model the association of falls and injurious falls
with data measured on patients. The forward stepwise selection procedure was
used to determine which patient measures were associated with falls and
injurious falls, and Bayesian model averaging (BMA) was used to account for
uncertainty in this variable selection procedure. Data on 99 patients for a
12-month time period were considered in this analysis. Fifty five percent of
the patients experienced at least one fall, with a total of 335 falls cases;
25% of which were injurious falls. Fearful, Tinetti gait, and previous falls
were the risk factors for fall/non-fall, with 77% accuracy, 76% sensitivity,
and 76% specificity. Fall time, body mass index, anxiety, balance, gait, and
gender were the risk factors associated with injurious falls. Thus, attaining
normal body mass index, improving balance and gait could be seen as preventive
efforts for injurious falls. There was no significant difference in the risk of
falls between males and females, yet if falls occurred, females were more
likely to get injured than males.
| stat |
Lecture Notes: Selected topics on robust statistical learning theory | These notes gather recent results on robust statistical learning theory. The
goal is to stress the main principles underlying the construction and
theoretical analysis of these estimators rather than provide an exhaustive
account on this rapidly growing field. The notes are the basis of lectures
given at the conference StatMathAppli 2019.
| stat |
A data-driven P-spline smoother and the P-Spline-GARCH-models | Penalized spline smoothing of time series and its asymptotic properties are
studied. A data-driven algorithm for selecting the smoothing parameter is
developed. The proposal is applied to define a semiparametric extension of the
well-known Spline-GARCH, called a P-Spline-GARCH, based on the log-data
transformation of the squared returns. It is shown that now the errors process
is exponentially strong mixing with finite moments of all orders. Asymptotic
normality of the P-spline smoother in this context is proved. Practical
relevance of the proposal is illustrated by data examples and simulation. The
proposal is further applied to value at risk and expected shortfall.
| stat |
Introduction to logistic regression | For random field theory based multiple comparison corrections In brain
imaging, it is often necessary to compute the distribution of the supremum of a
random field. Unfortunately, computing the distribution of the supremum of the
random field is not easy and requires satisfying many distributional
assumptions that may not be true in real data. Thus, there is a need to come up
with a different framework that does not use the traditional statistical
hypothesis testing paradigm that requires to compute p-values. With this as a
motivation, we can use a different approach called the logistic regression that
does not require computing the p-value and still be able to localize the
regions of brain network differences. Unlike other discriminant and
classification techniques that tried to classify preselected feature vectors,
the method here does not require any preselected feature vectors and performs
the classification at each edge level.
| stat |
Interpreting Missing Data Patterns in the ICU | PURPOSE: Clinical examinations are performed on the basis of necessity.
However, our decisions to investigate and document are influenced by various
other factors, such as workload and preconceptions. Data missingness patterns
may contain insights into conscious and unconscious norms of clinical practice.
METHODS: We examine data from the SPOTLIGHT study, a multi-centre cohort study
of the effect of prompt ICU admission on mortality. We identify missing values
and generate an auxiliary dataset indicating the missing entries. We deploy
sparse Gaussian Graphical modelling techniques to identify conditional
dependencies between the observed data and missingness patterns. We quantify
these associations with sparse partial correlation, correcting for multiple
collinearity. RESULTS: We identify 35 variables which significantly influence
data missingness patterns (alpha = 0.01). We identify reduced recording of
essential monitoring such as temperature (partial corr. = -0.0542, p =
6.65e-10), respiratory rate (partial corr. = -0.0437, p = 5.15e-07) and urea
(partial corr. = -0.0263, p = 0.001611) in patients with reduced consciousness.
We demonstrate reduction of temperature (partial corr. = -0.04, p = 8.5e-06),
urine output (partial corr. = -0.05, p = 7.5e-09), lactate (partial corr. =
-0.03, p = 0.00032) and bilirubin (partial corr. = -0.03, p = 0.00137)
monitoring due to winter pressures. We provide statistical evidence of Missing
Not at Random patterns in FiO2 and SF ratio recording. CONCLUSIONS: Graphical
missingness analysis offers valuable insights into critical care delivery,
identifying specific areas for quality improvement.
| stat |
LS-SVR as a Bayesian RBF network | We show theoretical similarities between the Least Squares Support Vector
Regression (LS-SVR) model with a Radial Basis Functions (RBF) kernel and
maximum a posteriori (MAP) inference on Bayesian RBF networks with a specific
Gaussian prior on the regression weights. Although previous works have pointed
out similar expressions between those learning approaches, we explicit and
formally state the existing correspondences. We empirically demonstrate our
result by performing computational experiments with standard regression
benchmarks. Our findings open a range of possibilities to improve LS-SVR by
borrowing strength from well-established developments in Bayesian methodology.
| stat |
Towards better healthcare: What could and should be automated? | While artificial intelligence (AI) and other automation technologies might
lead to enormous progress in healthcare, they may also have undesired
consequences for people working in the field. In this interdisciplinary study,
we capture empirical evidence of not only what healthcare work could be
automated, but also what should be automated. We quantitatively investigate
these research questions by utilizing probabilistic machine learning models
trained on thousands of ratings, provided by both healthcare practitioners and
automation experts. Based on our findings, we present an analytical tool
(Automatability-Desirability Matrix) to support policymakers and organizational
leaders in developing practical strategies on how to harness the positive power
of automation technologies, while accompanying change and empowering
stakeholders in a participatory fashion.
| stat |
Baseline Drift Estimation for Air Quality Data Using Quantile Trend
Filtering | We address the problem of estimating smoothly varying baseline trends in time
series data. This problem arises in a wide range of fields, including
chemistry, macroeconomics, and medicine; however, our study is motivated by the
analysis of data from low cost air quality sensors. Our methods extend the
quantile trend filtering framework to enable the estimation of multiple
quantile trends simultaneously while ensuring that the quantiles do not cross.
To handle the computational challenge posed by very long time series, we
propose a parallelizable alternating direction method of moments (ADMM)
algorithm. The ADMM algorthim enables the estimation of trends in a piecewise
manner, both reducing the computation time and extending the limits of the
method to larger data sizes. We also address smoothing parameter selection and
propose a modified criterion based on the extended Bayesian Information
Criterion. Through simulation studies and our motivating application to low
cost air quality sensor data, we demonstrate that our model provides better
quantile trend estimates than existing methods and improves signal
classification of low-cost air quality sensor output.
| stat |
Sinkhorn Natural Gradient for Generative Models | We consider the problem of minimizing a functional over a parametric family
of probability measures, where the parameterization is characterized via a
push-forward structure. An important application of this problem is in training
generative adversarial networks. In this regard, we propose a novel Sinkhorn
Natural Gradient (SiNG) algorithm which acts as a steepest descent method on
the probability space endowed with the Sinkhorn divergence. We show that the
Sinkhorn information matrix (SIM), a key component of SiNG, has an explicit
expression and can be evaluated accurately in complexity that scales
logarithmically with respect to the desired accuracy. This is in sharp contrast
to existing natural gradient methods that can only be carried out
approximately. Moreover, in practical applications when only Monte-Carlo type
integration is available, we design an empirical estimator for SIM and provide
the stability analysis. In our experiments, we quantitatively compare SiNG with
state-of-the-art SGD-type solvers on generative tasks to demonstrate its
efficiency and efficacy of our method.
| stat |
VAT tax gap prediction: a 2-steps Gradient Boosting approach | Tax evasion is the illegal evasion of taxes by individuals, corporations, and
trusts. The revenue loss from tax avoidance can undermine the effectiveness and
equity of the government policies. A standard measure of tax evasion is the tax
gap, that can be estimated as the difference between the total amounts of tax
theoretically collectable and the total amounts of tax actually collected in a
given period. This paper presents an original contribution to bottom-up
approach, based on results from fiscal audits, through the use of Machine
Learning. The major disadvantage of bottom-up approaches is represented by
selection bias when audited taxpayers are not randomly selected, as in the case
of audits performed by the Italian Revenue Agency. Our proposal, based on a
2-steps Gradient Boosting model, produces a robust tax gap estimate and, embeds
a solution to correct for the selection bias which do not require any
assumptions on the underlying data distribution. The 2-steps Gradient Boosting
approach is used to estimate the Italian Value-added tax (VAT) gap on
individual firms on the basis of fiscal and administrative data income tax
returns gathered from Tax Administration Data Base, for the fiscal year 2011.
The proposed method significantly boost the performance in predicting with
respect to the classical parametric approaches.
| stat |
Review of Probability Distributions for Modeling Count Data | Count data take on non-negative integer values and are challenging to
properly analyze using standard linear-Gaussian methods such as linear
regression and principal components analysis. Generalized linear models enable
direct modeling of counts in a regression context using distributions such as
the Poisson and negative binomial. When counts contain only relative
information, multinomial or Dirichlet-multinomial models can be more
appropriate. We review some of the fundamental connections between multinomial
and count models from probability theory, providing detailed proofs. These
relationships are useful for methods development in applications such as topic
modeling of text data and genomics.
| stat |
High-dimensional copula variational approximation through transformation | Variational methods are attractive for computing Bayesian inference for
highly parametrized models and large datasets where exact inference is
impractical. They approximate a target distribution - either the posterior or
an augmented posterior - using a simpler distribution that is selected to
balance accuracy with computational feasibility. Here we approximate an
element-wise parametric transformation of the target distribution as
multivariate Gaussian or skew-normal. Approximations of this kind are implicit
copula models for the original parameters, with a Gaussian or skew-normal
copula function and flexible parametric margins. A key observation is that
their adoption can improve the accuracy of variational inference in high
dimensions at limited or no additional computational cost. We consider the
Yeo-Johnson and G&H transformations, along with sparse factor structures for
the scale matrix of the Gaussian or skew-normal. We also show how to implement
efficient reparametrization gradient methods for these copula-based
approximations. The efficacy of the approach is illustrated by computing
posterior inference for three different models using six real datasets. In each
case, we show that our proposed copula model distributions are more accurate
variational approximations than Gaussian or skew-normal distributions, but at
only a minor or no increase in computational cost.
| stat |
Improved Generalization Bound of Group Invariant / Equivariant Deep
Networks via Quotient Feature Space | A large number of group invariant (or equivariant) networks have succeeded in
handling invariant data such as point clouds and graphs. However,
generalization theory for the networks has not been well developed, because
several essential factors for generalization theory, such as size and margin
distribution, are not very suitable to explain invariance and equivariance. In
this paper, we develop a generalization error bound for invariant and
equivariant deep neural networks. To describe the effect of the properties on
generalization, we develop a quotient feature space, which measures the effect
of group action for invariance or equivariance. Our main theorem proves that
the volume of quotient feature spaces largely improves the main term of the
developed bound. We apply our result to a specific invariant and equivariant
networks, such as DeepSets (Zaheer et al. (2017)), then show that their
generalization bound is drastically improved by $\sqrt{n!}$ where $n$ is a
number of permuting coordinates of data. Moreover, we additionally discuss the
representation power of invariant DNNs, and show that they can achieve an
optimal approximation rate. This paper is the first study to provide a general
and tight generalization bound for a broad class of group invariant and
equivariant deep neural networks.
| stat |
An adaptive multiclass nearest neighbor classifier | We consider a problem of multiclass classification, where the training sample
$S_n = \{(X_i, Y_i)\}_{i=1}^n$ is generated from the model $\mathbb P(Y = m | X
= x) = \eta_m(x)$, $1 \leq m \leq M$, and $\eta_1(x), \dots, \eta_M(x)$ are
unknown $\alpha$-Holder continuous functions.Given a test point $X$, our goal
is to predict its label. A widely used $\mathsf k$-nearest-neighbors classifier
constructs estimates of $\eta_1(X), \dots, \eta_M(X)$ and uses a plug-in rule
for the prediction. However, it requires a proper choice of the smoothing
parameter $\mathsf k$, which may become tricky in some situations. In our
solution, we fix several integers $n_1, \dots, n_K$, compute corresponding
$n_k$-nearest-neighbor estimates for each $m$ and each $n_k$ and apply an
aggregation procedure. We study an algorithm, which constructs a convex
combination of these estimates such that the aggregated estimate behaves
approximately as well as an oracle choice. We also provide a non-asymptotic
analysis of the procedure, prove its adaptation to the unknown smoothness
parameter $\alpha$ and to the margin and establish rates of convergence under
mild assumptions.
| stat |
Commuting Network Spillovers and COVID-19 Deaths Across US Counties | This study explored how population mobility flows form commuting networks
across US counties and influence the spread of COVID-19. We utilized 3-level
mixed effects negative binomial regression models to estimate the impact of
network COVID-19 exposure on county confirmed cases and deaths over time. We
also conducted weighting-based analyses to estimate the causal effect of
network exposure. Results showed that commuting networks matter for COVID-19
deaths and cases, net of spatial proximity, socioeconomic, and demographic
factors. Different local racial and ethnic concentrations are also associated
with unequal outcomes. These findings suggest that commuting is an important
causal mechanism in the spread of COVID-19 and highlight the significance of
interconnected of communities. The results suggest that local level mitigation
and prevention efforts are more effective when complemented by similar efforts
in the network of connected places. Implications for research on inequality in
health and flexible work arrangements are discussed.
| stat |
Hybrid Kronecker Product Decomposition and Approximation | Discovering the underlying low dimensional structure of high dimensional data
has attracted a significant amount of researches recently and has shown to have
a wide range of applications. As an effective dimension reduction tool,
singular value decomposition is often used to analyze high dimensional
matrices, which are traditionally assumed to have a low rank matrix
approximation. In this paper, we propose a new approach. We assume a high
dimensional matrix can be approximated by a sum of a small number of Kronecker
products of matrices with potentially different configurations, named as a
hybird Kronecker outer Product Approximation (hKoPA). It provides an extremely
flexible way of dimension reduction compared to the low-rank matrix
approximation. Challenges arise in estimating a hKoPA when the configurations
of component Kronecker products are different or unknown. We propose an
estimation procedure when the set of configurations are given and a joint
configuration determination and component estimation procedure when the
configurations are unknown. Specifically, a least squares backfitting algorithm
is used when the configuration is given. When the configuration is unknown, an
iterative greedy algorithm is used. Both simulation and real image examples
show that the proposed algorithms have promising performances. The hybrid
Kronecker product approximation may have potentially wider applications in low
dimensional representation of high dimensional data
| stat |
Deep Generative Markov State Models | We propose a deep generative Markov State Model (DeepGenMSM) learning
framework for inference of metastable dynamical systems and prediction of
trajectories. After unsupervised training on time series data, the model
contains (i) a probabilistic encoder that maps from high-dimensional
configuration space to a small-sized vector indicating the membership to
metastable (long-lived) states, (ii) a Markov chain that governs the
transitions between metastable states and facilitates analysis of the long-time
dynamics, and (iii) a generative part that samples the conditional distribution
of configurations in the next time step. The model can be operated in a
recursive fashion to generate trajectories to predict the system evolution from
a defined starting state and propose new configurations. The DeepGenMSM is
demonstrated to provide accurate estimates of the long-time kinetics and
generate valid distributions for molecular dynamics (MD) benchmark systems.
Remarkably, we show that DeepGenMSMs are able to make long time-steps in
molecular configuration space and generate physically realistic structures in
regions that were not seen in training data.
| stat |
Human mobility and COVID-19 initial dynamics | Mobility data at EU scale can help understand the dynamics of the pandemic
and possibly limit the impact of future waves. Still, since a reliable and
consistent method to measure the evolution of contagion at international level
is missing, a systematic analysis of the relationship between human mobility
and virus spread has never been conducted. A notable exceptions are France and
Italy, for which data on excess deaths, an indirect indicator which is
generally considered to be less affected by national and regional assumptions,
are available at department and municipality level, respectively. Using this
information together with anonymised and aggregated mobile data, this study
shows that mobility alone can explain up to 92% of the initial spread in these
two EU countries, while it has a slow decay effect after lockdown measures,
meaning that mobility restrictions seem to have effectively contribute to save
lives. It also emerges that internal mobility is more important than mobility
across provinces and that the typical lagged positive effect of reduced human
mobility on reducing excess deaths is around 14-20 days. An analogous analysis
relative to Spain, for which an IgG SARS-Cov-2 antibody screening study at
province level is used instead of excess deaths statistics, confirms the
findings. The same approach adopted in this study can be easily extended to
other European countries, as soon as reliable data on the spreading of the
virus at a suitable level of granularity will be available. Looking at past
data, relative to the initial phase of the outbreak in EU Member States, this
study shows in which extent the spreading of the virus and human mobility are
connected.
| stat |
Clustering Airbnb Reviews | In the last decade, online customer reviews increasingly exert influence on
consumers' decision when booking accommodation online. The renewal importance
to the concept of word-of mouth is reflected in the growing interests in
investigating consumers' experience by analyzing their online reviews through
the process of text mining and sentiment analysis. A clustering approach is
developed for Boston Airbnb reviews submitted in the English language and
collected from 2009 to 2016. This approach is based on a mixture of latent
variable models, which provides an appealing framework for handling clustered
binary data. We address here the problem of discovering meaningful segments of
consumers that are coherent from both the underlying topics and the sentiment
behind the reviews. A penalized mixture of latent traits approach is developed
to reduce the number of parameters and identify variables that are not
informative for clustering. The introduction of component-specific rate
parameters avoids the over-penalization that can occur when inferring a shared
rate parameter on clustered data. We divided the guests into four groups --
property driven guests, host driven guests, guests with recent overall negative
stay and guests with some negative experiences.
| stat |
STORM: Foundations of End-to-End Empirical Risk Minimization on the Edge | Empirical risk minimization is perhaps the most influential idea in
statistical learning, with applications to nearly all scientific and technical
domains in the form of regression and classification models. To analyze massive
streaming datasets in distributed computing environments, practitioners
increasingly prefer to deploy regression models on edge rather than in the
cloud. By keeping data on edge devices, we minimize the energy, communication,
and data security risk associated with the model. Although it is equally
advantageous to train models at the edge, a common assumption is that the model
was originally trained in the cloud, since training typically requires
substantial computation and memory. To this end, we propose STORM, an online
sketch for empirical risk minimization. STORM compresses a data stream into a
tiny array of integer counters. This sketch is sufficient to estimate a variety
of surrogate losses over the original dataset. We provide rigorous theoretical
analysis and show that STORM can estimate a carefully chosen surrogate loss for
the least-squares objective. In an exhaustive experimental comparison for
linear regression models on real-world datasets, we find that STORM allows
accurate regression models to be trained.
| stat |
Large-scale Heteroscedastic Regression via Gaussian Process | Heteroscedastic regression considering the varying noises among observations
has many applications in the fields like machine learning and statistics. Here
we focus on the heteroscedastic Gaussian process (HGP) regression which
integrates the latent function and the noise function together in a unified
non-parametric Bayesian framework. Though showing remarkable performance, HGP
suffers from the cubic time complexity, which strictly limits its application
to big data. To improve the scalability, we first develop a variational sparse
inference algorithm, named VSHGP, to handle large-scale datasets. Furthermore,
two variants are developed to improve the scalability and capability of VSHGP.
The first is stochastic VSHGP (SVSHGP) which derives a factorized evidence
lower bound, thus enhancing efficient stochastic variational inference. The
second is distributed VSHGP (DVSHGP) which (i) follows the Bayesian committee
machine formalism to distribute computations over multiple local VSHGP experts
with many inducing points; and (ii) adopts hybrid parameters for experts to
guard against over-fitting and capture local variety. The superiority of DVSHGP
and SVSHGP as compared to existing scalable heteroscedastic/homoscedastic GPs
is then extensively verified on various datasets.
| stat |
Broadcasted Nonparametric Tensor Regression | We propose a novel broadcasting idea to model the nonlinearity in tensor
regression non-parametrically. Unlike existing non-parametric tensor regression
models, the resulting model strikes a good balance between flexibility and
interpretability. A penalized estimation and corresponding algorithm are
proposed. Our theoretical investigation, which allows the dimensions of the
tensor covariate to diverge, indicates that the proposed estimation enjoys
desirable convergence rate. Numerical experiments are conducted to confirm the
theoretical finding and show that the proposed model has advantage over
existing linear counterparts.
| stat |
Musings about Constructions of Efficient Latin Hypercube Designs with
Flexible Run-sizes | Efficient Latin hypercube designs (LHDs), including maximin distance LHDs,
maximum projection LHDs and orthogonal LHDs, are widely used in computer
experiments. It is challenging to construct such designs with flexible sizes,
especially for large ones. In the current literature, various algebraic methods
and search algorithms have been proposed for identifying efficient LHDs, each
having its own pros and cons. In this paper, we review, summarize and compare
some currently popular methods aiming to provide guidance for experimenters on
what method should be used in practice. Using the R package we developed which
integrates and improves various algebraic and searching methods, many of the
designs found in this paper are better than the existing ones. They are easy to
use for practitioners and can serve as benchmarks for the future developments
on LHDs.
| stat |
Where Does Haydn End and Mozart Begin? Composer Classification of String
Quartets | For centuries, the history and music of Joseph Franz Haydn and Wolfgang
Amadeus Mozart have been compared by scholars. Recently, the growing field of
music information retrieval (MIR) has offered quantitative analyses to
complement traditional qualitative analyses of these composers. In this MIR
study, we classify the composer of Haydn and Mozart string quartets based on
the content of their scores. Our contribution is an interpretable statistical
and machine learning approach that provides high classification accuracies and
musical relevance. We develop novel global features that are automatically
computed from symbolic data and informed by musicological Haydn-Mozart
comparative studies, particularly relating to the sonata form. Several of these
proposed features are found to be important for distinguishing between Haydn
and Mozart string quartets. Our Bayesian logistic regression model attains
leave-one-out classification accuracies over 84%, higher than prior works and
providing interpretations that could aid in assessing musicological claims.
Overall, our work can help expand the longstanding dialogue surrounding Haydn
and Mozart and exemplify the benefit of interpretable machine learning in MIR,
with potential applications to music generation and classification of other
classical composers.
| stat |
Learning to Generate Fair Clusters from Demonstrations | Fair clustering is the process of grouping similar entities together, while
satisfying a mathematically well-defined fairness metric as a constraint. Due
to the practical challenges in precise model specification, the prescribed
fairness constraints are often incomplete and act as proxies to the intended
fairness requirement, leading to biased outcomes when the system is deployed.
We examine how to identify the intended fairness constraint for a problem based
on limited demonstrations from an expert. Each demonstration is a clustering
over a subset of the data.
We present an algorithm to identify the fairness metric from demonstrations
and generate clusters using existing off-the-shelf clustering techniques, and
analyze its theoretical properties. To extend our approach to novel fairness
metrics for which clustering algorithms do not currently exist, we present a
greedy method for clustering. Additionally, we investigate how to generate
interpretable solutions using our approach. Empirical evaluation on three
real-world datasets demonstrates the effectiveness of our approach in quickly
identifying the underlying fairness and interpretability constraints, which are
then used to generate fair and interpretable clusters.
| stat |
A spatio-temporal model to understand forest fires causality in Europe | Forest fires are the outcome of a complex interaction between environmental
factors, topography and socioeconomic factors (Bedia et al, 2014). Therefore,
understand causality and early prediction are crucial elements for controlling
such phenomenon and saving lives.The aim of this study is to build
spatio-temporal model to understand causality of forest fires in Europe, at
NUTS2 level between 2012 and 2016, using environmental and socioeconomic
variables.We have considered a disease mapping approach, commonly used in small
area studies to assess thespatial pattern and to identify areas characterised
by unusually high or low relative risk.
| stat |
Banach Space Representer Theorems for Neural Networks and Ridge Splines | We develop a variational framework to understand the properties of the
functions learned by neural networks fit to data. We propose and study a family
of continuous-domain linear inverse problems with total variation-like
regularization in the Radon domain subject to data fitting constraints. We
derive a representer theorem showing that finite-width, single-hidden layer
neural networks are solutions to these inverse problems. We draw on many
techniques from variational spline theory and so we propose the notion of
polynomial ridge splines, which correspond to single-hidden layer neural
networks with truncated power functions as the activation function. The
representer theorem is reminiscent of the classical reproducing kernel Hilbert
space representer theorem, but we show that the neural network problem is posed
over a non-Hilbertian Banach space. While the learning problems are posed in
the continuous-domain, similar to kernel methods, the problems can be recast as
finite-dimensional neural network training problems. These neural network
training problems have regularizers which are related to the well-known weight
decay and path-norm regularizers. Thus, our result gives insight into
functional characteristics of trained neural networks and also into the design
neural network regularizers. We also show that these regularizers promote
neural network solutions with desirable generalization properties.
| stat |
Adaptive and optimal online linear regression on $\ell^1$-balls | We consider the problem of online linear regression on individual sequences.
The goal in this paper is for the forecaster to output sequential predictions
which are, after $T$ time rounds, almost as good as the ones output by the best
linear predictor in a given $\ell^1$-ball in $\\R^d$. We consider both the
cases where the dimension~$d$ is small and large relative to the time horizon
$T$. We first present regret bounds with optimal dependencies on $d$, $T$, and
on the sizes $U$, $X$ and $Y$ of the $\ell^1$-ball, the input data and the
observations. The minimax regret is shown to exhibit a regime transition around
the point $d = \sqrt{T} U X / (2 Y)$. Furthermore, we present efficient
algorithms that are adaptive, \ie, that do not require the knowledge of $U$,
$X$, $Y$, and $T$, but still achieve nearly optimal regret bounds.
| stat |
Additive multivariate Gaussian processes for joint species distribution
modeling with heterogeneous data | Species distribution models (SDM) are a key tool in ecology, conservation and
management of natural resources. Two key components of the state-of-the-art
SDMs are the description for species distribution response along environmental
covariates and the spatial random effect. Joint species distribution models
(JSDMs) additionally include interspecific correlations which have been shown
to improve their descriptive and predictive performance compared to single
species models. Current JSDMs are restricted to hierarchical generalized linear
modeling framework. These parametric models have trouble in explaining changes
in abundance due, e.g., highly non-linear physical tolerance limits which is
particularly important when predicting species distribution in new areas or
under scenarios of environmental change. On the other hand, semi-parametric
response functions have been shown to improve the predictive performance of
SDMs in these tasks in single species models. Here, we propose JSDMs where the
responses to environmental covariates are modeled with additive multivariate
Gaussian processes coded as linear models of coregionalization. These allow
inference for wide range of functional forms and interspecific correlations
between the responses. We propose also an efficient approach for inference with
Laplace approximation and parameterization of the interspecific covariance
matrices on the euclidean space. We demonstrate the benefits of our model with
two small scale examples and one real world case study. We use cross-validation
to compare the proposed model to analogous semi-parametric single species
models and parametric single and joint species models in interpolation and
extrapolation tasks. The proposed model outperforms the alternative models in
all cases. We also show that the proposed model can be seen as an extension of
the current state-of-the-art JSDMs to semi-parametric models.
| stat |
Subsets and Splits
No saved queries yet
Save your SQL queries to embed, download, and access them later. Queries will appear here once saved.