title
stringlengths 6
244
| abstract
stringlengths 19
5.09k
| label
stringclasses 10
values |
|---|---|---|
Causal Inference in Higher Education: Building Better Curriculums | Higher educational institutions constantly look for ways to meet students'
needs and support them through graduation. Recent work in the field of learning
analytics have developed methods for grade prediction and course
recommendations. Although these methods work well, they often fail to discover
causal relationships between courses, which may not be evident through
correlation-based methods. In this work, we aim at understanding the causal
relationships between courses to aid universities in designing better academic
pathways for students and to help them make better choices. Our methodology
employs methods of causal inference to study these relationships using
historical student performance data. We make use of a doubly-robust method of
matching and regression in order to obtain the casual relationship between a
pair of courses. The results were validated by the existing prerequisite
structure and by cross-validation of the regression model. Further, our
approach was also tested for robustness and sensitivity to certain
hyperparameters. This methodology shows promising results and is a step forward
towards building better academic pathways for students.
| stat |
Assessing the effect of advertising expenditures upon sales: a Bayesian
structural time series model | We propose a robust implementation of the Nerlove--Arrow model using a
Bayesian structural time series model to explain the relationship between
advertising expenditures of a country-wide fast-food franchise network with its
weekly sales. Thanks to the flexibility and modularity of the model, it is well
suited to generalization to other markets or situations. Its Bayesian nature
facilitates incorporating \emph{a priori} information (the manager's views),
which can be updated with relevant data. This aspect of the model will be used
to present a strategy of budget scheduling across time and channels.
| stat |
Multiscale regression on unknown manifolds | We consider the regression problem of estimating functions on $\mathbb{R}^D$
but supported on a $d$-dimensional manifold $ \mathcal{M} \subset \mathbb{R}^D
$ with $ d \ll D $. Drawing ideas from multi-resolution analysis and nonlinear
approximation, we construct low-dimensional coordinates on $\mathcal{M}$ at
multiple scales, and perform multiscale regression by local polynomial fitting.
We propose a data-driven wavelet thresholding scheme that automatically adapts
to the unknown regularity of the function, allowing for efficient estimation of
functions exhibiting nonuniform regularity at different locations and scales.
We analyze the generalization error of our method by proving finite sample
bounds in high probability on rich classes of priors. Our estimator attains
optimal learning rates (up to logarithmic factors) as if the function was
defined on a known Euclidean domain of dimension $d$, instead of an unknown
manifold embedded in $\mathbb{R}^D$. The implemented algorithm has quasilinear
complexity in the sample size, with constants linear in $D$ and exponential in
$d$. Our work therefore establishes a new framework for regression on
low-dimensional sets embedded in high dimensions, with fast implementation and
strong theoretical guarantees.
| stat |
Bayesian Data Synthesis and Disclosure Risk Quantification: An
Application to the Consumer Expenditure Surveys | The release of synthetic data generated from a model estimated on the data
helps statistical agencies disseminate respondent-level data with high utility
and privacy protection. Motivated by the challenge of disseminating sensitive
variables containing geographic information in the Consumer Expenditure Surveys
(CE) at the U.S. Bureau of Labor Statistics, we propose two non-parametric
Bayesian models as data synthesizers for the county identifier of each data
record: a Bayesian latent class model and a Bayesian areal model. Both data
synthesizers use Dirichlet Process priors to cluster observations of similar
characteristics and allow borrowing information across observations. We develop
innovative disclosure risks measures to quantify inherent risks in the
confidential CE data and how those data risks are ameliorated by our proposed
synthesizers. By creating a lower bound and an upper bound of disclosure risks
under a minimum and a maximum disclosure risks scenarios respectively, our
proposed inherent risks measures provide a range of acceptable disclosure risks
for evaluating risks level in the synthetic datasets.
| stat |
Convergence rates for optimised adaptive importance samplers | Adaptive importance samplers are adaptive Monte Carlo algorithms to estimate
expectations with respect to some target distribution which \textit{adapt}
themselves to obtain better estimators over a sequence of iterations. Although
it is straightforward to show that they have the same $\mathcal{O}(1/\sqrt{N})$
convergence rate as standard importance samplers, where $N$ is the number of
Monte Carlo samples, the behaviour of adaptive importance samplers over the
number of iterations has been left relatively unexplored. In this work, we
investigate an adaptation strategy based on convex optimisation which leads to
a class of adaptive importance samplers termed \textit{optimised adaptive
importance samplers} (OAIS). These samplers rely on the iterative minimisation
of the $\chi^2$-divergence between an exponential-family proposal and the
target. The analysed algorithms are closely related to the class of adaptive
importance samplers which minimise the variance of the weight function. We
first prove non-asymptotic error bounds for the mean squared errors (MSEs) of
these algorithms, which explicitly depend on the number of iterations and the
number of samples together. The non-asymptotic bounds derived in this paper
imply that when the target belongs to the exponential family, the $L_2$ errors
of the optimised samplers converge to the optimal rate of
$\mathcal{O}(1/\sqrt{N})$ and the rate of convergence in the number of
iterations are explicitly provided. When the target does not belong to the
exponential family, the rate of convergence is the same but the asymptotic
$L_2$ error increases by a factor $\sqrt{\rho^\star} > 1$, where $\rho^\star -
1$ is the minimum $\chi^2$-divergence between the target and an
exponential-family proposal.
| stat |
Spatially Aggregated Gaussian Processes with Multivariate Areal Outputs | We propose a probabilistic model for inferring the multivariate function from
multiple areal data sets with various granularities. Here, the areal data are
observed not at location points but at regions. Existing regression-based
models can only utilize the sufficiently fine-grained auxiliary data sets on
the same domain (e.g., a city). With the proposed model, the functions for
respective areal data sets are assumed to be a multivariate dependent Gaussian
process (GP) that is modeled as a linear mixing of independent latent GPs.
Sharing of latent GPs across multiple areal data sets allows us to effectively
estimate the spatial correlation for each areal data set; moreover it can
easily be extended to transfer learning across multiple domains. To handle the
multivariate areal data, we design an observation model with a spatial
aggregation process for each areal data set, which is an integral of the mixed
GP over the corresponding region. By deriving the posterior GP, we can predict
the data value at any location point by considering the spatial correlations
and the dependences between areal data sets, simultaneously. Our experiments on
real-world data sets demonstrate that our model can 1) accurately refine
coarse-grained areal data, and 2) offer performance improvements by using the
areal data sets from multiple domains.
| stat |
An Advanced Hidden Markov Model for Hourly Rainfall Time Series | For hydrological applications, such as urban flood modelling, it is often
important to be able to simulate sub-daily rainfall time series from stochastic
models. However, modelling rainfall at this resolution poses several
challenges, including a complex temporal structure including long dry periods,
seasonal variation in both the occurrence and intensity of rainfall, and
extreme values.
We illustrate how the hidden Markov framework can be adapted to construct a
compelling model for sub-daily rainfall, which is capable of capturing all of
these important characteristics well. These adaptations include clone states
and non-stationarity in both the transition matrix and conditional models. Set
in the Bayesian framework, a rich quantification of both parametric and
predictive uncertainty is available, and thorough model checking is made
possible through posterior predictive analyses. Results from the model are
interpretable, allowing for meaningful examination of seasonal variation and
medium to long term trends in rainfall occurrence and intensity. To demonstrate
the effectiveness of our approach, both in terms of model fit and
interpretability, we apply the model to an 8-year long time series of hourly
observations.
| stat |
Patient-Specific Effects of Medication Using Latent Force Models with
Gaussian Processes | Multi-output Gaussian processes (GPs) are a flexible Bayesian nonparametric
framework that has proven useful in jointly modeling the physiological states
of patients in medical time series data. However, capturing the short-term
effects of drugs and therapeutic interventions on patient physiological state
remains challenging. We propose a novel approach that models the effect of
interventions as a hybrid Gaussian process composed of a GP capturing patient
physiology convolved with a latent force model capturing effects of treatments
on specific physiological features. This convolution of a multi-output GP with
a GP including a causal time-marked kernel leads to a well-characterized model
of the patients' physiological state responding to interventions. We show that
our model leads to analytically tractable cross-covariance functions, allowing
scalable inference. Our hierarchical model includes estimates of
patient-specific effects but allows sharing of support across patients. Our
approach achieves competitive predictive performance on challenging hospital
data, where we recover patient-specific response to the administration of three
common drugs: one antihypertensive drug and two anticoagulants.
| stat |
Contagion Dynamics for Manifold Learning | Contagion maps exploit activation times in threshold contagions to assign
vectors in high-dimensional Euclidean space to the nodes of a network. A point
cloud that is the image of a contagion map reflects both the structure
underlying the network and the spreading behaviour of the contagion on it.
Intuitively, such a point cloud exhibits features of the network's underlying
structure if the contagion spreads along that structure, an observation which
suggests contagion maps as a viable manifold-learning technique. We test
contagion maps as a manifold-learning tool on a number of different real-world
and synthetic data sets, and we compare their performance to that of Isomap,
one of the most well-known manifold-learning algorithms. We find that, under
certain conditions, contagion maps are able to reliably detect underlying
manifold structure in noisy data, while Isomap fails due to noise-induced
error. This consolidates contagion maps as a technique for manifold learning.
| stat |
Bayesian Modeling of Spatial Transcriptomics Data via a Modified Ising
Model | Recent technology breakthroughs in spatial molecular profiling (SMP), such as
spatial transcriptomics sequencing, have enabled the comprehensive molecular
characterization of single cells while preserving spatial and morphological
information. One immediate question is how to identify spatially variable (SV)
genes. Most of the current work builds upon the geostatistical model with
Gaussian process that relies on the selection of \textit{ad hoc} kernels to
account for spatial expression patterns. To overcome this potential challenge
and capture more spatial patterns, we introduced a Bayesian modeling framework
to identify SV genes. Our model first dichotomized the complex sequencing count
data into latent binary gene expression levels. Then, binary pattern
quantification problem is considered as a spatial correlation estimation
problem via a modified Ising model using Hamiltonian energy to characterize
spatial patterns. We used auxiliary variable Markov chain Monte Carlo
algorithms to sample from the posterior distribution with an intractable
normalizing constant. Simulation results showed high accuracy in detecting SV
genes compared with kernel-based alternatives. We also applied our model to two
real datasets and discovered novel spatial patterns that shed light on the
biological mechanisms. This statistical methodology presents a new perspective
for characterizing spatial patterns from SMP data.
| stat |
Collective dynamics of pedestrians in a non-panic evacuation scenario | We present a study of pedestrian motion along a corridor in a non-panic
regime (e.g., schools, hospitals or airports). Such situations have been
discussed so far within the Social Force Model (SFM). We suggest to enrich this
model by interactions based on the velocity of the particles and some
randomness, both of which we introduce using the ideas of the Vicsek Model
(VM). This new model allows to introduce fluctuations for a given average speed
and geometry, and considering that the alignment interactions are modulated by
an external control parameter (the noise $\eta$) allows to introduce phase
transitions between ordered and disordered states. We have compared simulations
of pedestrian motion along a corridor using (a) the VM with two boundary
conditions (periodic and bouncing back) and with or without desired direction
of motion, (b) the SFM, and (c) the new model SFM+VM. The study of steady-state
configurations in the VM with confined geometry shows the expected bands
perpendicular to the motion direction, while in the SFM and SFM+VM particles
order in stripes of a given width $w$ along the direction of motion. The
results in the SFM+VM case show that $w(t)\simeq t^\alpha$ has a diffusive-like
behavior at low noise $\eta$ (dynamic exponent $\alpha \approx 1/2$), while it
is sub-diffusive at high values of external noise ($\alpha \approx 1/4$). We
observe the order-disorder transition in the VM with both boundary conditions,
but the application of a desired direction condition inhibits the existence of
disorder as expected. For the SFM+VM case we find a susceptibility maximum
which increases with system size as a function of noise strength indicative of
a order-disorder transition in the whole range of densities and speeds studied.
From our results we conclude that the new SFM+VM model is a well-suited model
to describe non-panic evacuation with diverse degrees of disorder.
| stat |
Conditional separable effects | Researchers are often interested in treatment effects on outcomes that are
only defined conditional on a post-treatment event status. For example, in a
study of the effect of different cancer treatments on quality of life at end of
follow-up, the quality of life of individuals who die during the study is
undefined. In these settings, a naive contrast of outcomes conditional on the
post-treatment variable is not an average causal effect, even in a randomized
experiment. Therefore the effect in the principal stratum of those who would
have the same value of the post-treatment variable regardless of treatment,
such as the always survivors in a truncation by death setting, is often
advocated for causal inference. While this principal stratum effect is a well
defined causal contrast, it is often hard to justify that it is relevant to
scientists, patients or policy makers, and it cannot be identified without
relying on unfalsifiable assumptions. Here we formulate alternative estimands,
the conditional separable effects, that have a natural causal interpretation
under assumptions that can be falsified in a randomized experiment. We provide
identification results and introduce different estimators, including a doubly
robust estimator derived from the nonparametric influence function. As an
illustration, we estimate a conditional separable effect of chemotherapies on
quality of life in patients with prostate cancer, using data from a randomized
clinical trial.
| stat |
Robust Inference for Partially Observed Functional Response Data | Irregular functional data in which densely sampled curves are observed over
different ranges pose a challenge for modeling and inference, and sensitivity
to outlier curves is a concern in applications. Motivated by applications in
quantitative ultrasound signal analysis, this paper investigates a class of
robust M-estimators for partially observed functional data including functional
location and quantile estimators. Consistency of the estimators is established
under general conditions on the partial observation process. Under smoothness
conditions on the class of M-estimators, asymptotic Gaussian process
approximations are established and used for large sample inference. The large
sample approximations justify a bootstrap approximation for robust inferences
about the functional response process. The performance is demonstrated in
simulations and in the analysis of irregular functional data from quantitative
ultrasound analysis.
| stat |
DeepHazard: neural network for time-varying risks | Prognostic models in survival analysis are aimed at understanding the
relationship between patients' covariates and the distribution of survival
time. Traditionally, semi-parametric models, such as the Cox model, have been
assumed. These often rely on strong proportionality assumptions of the hazard
that might be violated in practice. Moreover, they do not often include
covariate information updated over time. We propose a new flexible method for
survival prediction: DeepHazard, a neural network for time-varying risks. Our
approach is tailored for a wide range of continuous hazards forms, with the
only restriction of being additive in time. A flexible implementation, allowing
different optimization methods, along with any norm penalty, is developed.
Numerical examples illustrate that our approach outperforms existing
state-of-the-art methodology in terms of predictive capability evaluated
through the C-index metric. The same is revealed on the popular real datasets
as METABRIC, GBSG, and ACTG.
| stat |
The Estimation of Causal Effects of Multiple Treatments in Observational
Studies Using Bayesian Additive Regression Trees | There is currently a dearth of appropriate methods to estimate the causal
effects of multiple treatments when the outcome is binary. For such settings,
we propose the use of nonparametric Bayesian modeling, Bayesian Additive
Regression Trees (BART). We conduct an extensive simulation study to compare
BART to several existing, propensity score-based methods and to identify its
operating characteristics when estimating average treatment effects on the
treated. BART consistently demonstrates low bias and mean-squared errors. We
illustrate the use of BART through a comparative effectiveness analysis of a
large dataset, drawn from the latest SEER-Medicare linkage, on patients who
were operated via robotic-assisted surgery, video-assisted thoratic surgery or
open thoracotomy.
| stat |
Heavy-tailed distribution for combining dependent $p$-values with
asymptotic robustness | The issue of combining individual $p$-values to aggregate multiple small
effects is prevalent in many scientific investigations and is a long-standing
statistical topic. Many classical methods are designed for combining
independent and frequent signals in a traditional meta-analysis sense using the
sum of transformed $p$-values with the transformation of light-tailed
distributions, in which Fisher's method and Stouffer's method are the most
well-known. Since the early 2000, advances in big data promoted methods to
aggregate independent, sparse and weak signals, such as the renowned higher
criticism and Berk-Jones tests. Recently, Liu and Xie(2020) and Wilson(2019)
independently proposed Cauchy and harmonic mean combination tests to robustly
combine $p$-values under "arbitrary" dependency structure, where a notable
application is to combine $p$-values from a set of often correlated SNPs in
genome-wide association studies. The proposed tests are the transformation of
heavy-tailed distributions for improved power with the sparse signal. It calls
for a natural question to investigate heavy-tailed distribution transformation,
to understand the connection among existing methods, and to explore the
conditions for a method to possess robustness to dependency. In this paper, we
investigate the regularly varying distribution, which is a rich family of
heavy-tailed distribution and includes Pareto distribution as a special case.
We show that only an equivalent class of Cauchy and harmonic mean tests have
sufficient robustness to dependency in a practical sense. We also show an issue
caused by large negative penalty in the Cauchy method and propose a simple, yet
practical modification. Finally, we present simulations and apply to a
neuroticism GWAS application to verify the discovered theoretical insights and
provide practical guidance.
| stat |
Causal coupling inference from multivariate time series based on ordinal
partition transition networks | Identifying causal relationships is a challenging yet a crucial problem in
many fields of science like epidemiology, climatology, ecology, genomics,
economics and neuroscience, to mention only a few. Recent studies have
demonstrated that ordinal partition transition networks (OPTNs) allow to infer
the coupling direction between two dynamical systems. In this work, we
generalize this concept to the interaction between multiple dynamical systems
and propose a new method to detect causality in multivariate observational
data. We demonstrate that our approach can reliably identify the direction of
interaction and the corresponding delays with numerical simulations using
linear stochastic systems as well as nonlinear dynamical systems such as a
network of neural mass models. Finally, we apply our method to real-world
observational microelectrode array data from rodent brain slices to study the
causal effect networks underlying epileptic activity. Our results from
simulations as well as real-world data suggest that OPTNs can provide a
complementary approach to reliably infer causal effect networks from
multivariate observational data.
| stat |
Double Generative Adversarial Networks for Conditional Independence
Testing | In this article, we consider the problem of high-dimensional conditional
independence testing, which is a key building block in statistics and machine
learning. We propose a double generative adversarial networks (GANs)-based
inference procedure. We first introduce a double GANs framework to learn two
generators, and integrate the two generators to construct a doubly-robust test
statistic. We next consider multiple generalized covariance measures, and take
their maximum as our test statistic. Finally, we obtain the empirical
distribution of our test statistic through multiplier bootstrap. We show that
our test controls type-I error, while the power approaches one asymptotically.
More importantly, these theoretical guarantees are obtained under much weaker
and practically more feasible conditions compared to existing tests. We
demonstrate the efficacy of our test through both synthetic and real datasets.
| stat |
Matrix completion and extrapolation via kernel regression | Matrix completion and extrapolation (MCEX) are dealt with here over
reproducing kernel Hilbert spaces (RKHSs) in order to account for prior
information present in the available data. Aiming at a faster and
low-complexity solver, the task is formulated as a kernel ridge regression. The
resultant MCEX algorithm can also afford online implementation, while the class
of kernel functions also encompasses several existing approaches to MC with
prior information. Numerical tests on synthetic and real datasets show that the
novel approach performs faster than widespread methods such as alternating
least squares (ALS) or stochastic gradient descent (SGD), and that the recovery
error is reduced, especially when dealing with noisy data.
| stat |
Efficient Bayesian Modeling of Binary and Categorical Data in R: The UPG
Package | We introduce the UPG package for highly efficient Bayesian inference in
probit, logit, multinomial logit and binomial logit models. UPG offers a
convenient estimation framework for balanced and imbalanced data settings where
sampling efficiency is ensured through Markov chain Monte Carlo boosting
methods. All sampling algorithms are implemented in C++, allowing for rapid
parameter estimation. In addition, UPG provides several methods for fast
production of output tables and summary plots that are easily accessible to a
broad range of users.
| stat |
Ordinal Bayesian Optimisation | Bayesian optimisation is a powerful tool to solve expensive black-box
problems, but fails when the stationary assumption made on the objective
function is strongly violated, which is the case in particular for
ill-conditioned or discontinuous objectives. We tackle this problem by
proposing a new Bayesian optimisation framework that only considers the
ordering of variables, both in the input and output spaces, to fit a Gaussian
process in a latent space. By doing so, our approach is agnostic to the
original metrics on the original spaces. We propose two algorithms,
respectively based on an optimistic strategy and on Thompson sampling. For the
optimistic strategy we prove an optimal performance under the measure of regret
in the latent space. We illustrate the capability of our framework on several
challenging toy problems.
| stat |
Active learning for enumerating local minima based on Gaussian process
derivatives | We study active learning (AL) based on Gaussian Processes (GPs) for
efficiently enumerating all of the local minimum solutions of a black-box
function. This problem is challenging due to the fact that local solutions are
characterized by their zero gradient and positive-definite Hessian properties,
but those derivatives cannot be directly observed. We propose a new AL method
in which the input points are sequentially selected such that the confidence
intervals of the GP derivatives are effectively updated for enumerating local
minimum solutions. We theoretically analyze the proposed method and demonstrate
its usefulness through numerical experiments.
| stat |
On the usage of randomized p-values in the Schweder-Spjotvoll estimator | We are concerned with multiple test problems with composite null hypotheses
and the estimation of the proportion $\pi_{0}$ of true null hypotheses. The
Schweder-Spj\o tvoll estimator $\hat{\pi}_0$ utilizes marginal $p$-values and
only works properly if the $p$-values that correspond to the true null
hypotheses are uniformly distributed on $[0,1]$
($\mathrm{Uni}[0,1]$-distributed). In the case of composite null hypotheses,
marginal $p$-values are usually computed under least favorable parameter
configurations (LFCs). Thus, they are stochastically larger than
$\mathrm{Uni}[0,1]$ under non-LFCs in the null hypotheses. When using these
LFC-based $p$-values, $\hat{\pi}_0$ tends to overestimate $\pi_{0}$. We
introduce a new way of randomizing $p$-values that depends on a tuning
parameter $c\in[0,1]$, such that $c=0$ and $c=1$ lead to
$\mathrm{Uni}[0,1]$-distributed $p$-values, which are independent of the data,
and to the original LFC-based $p$-values, respectively. For a certain value
$c=c^{\star}$ the bias of $\hat{\pi}_0$ is minimized when using our randomized
$p$-values. This often also entails a smaller mean squared error of the
estimator as compared to the usage of the LFC-based $p$-values. We analyze
these points theoretically, and we demonstrate them numerically in computer
simulations under various standard statistical models.
| stat |
VCBART: Bayesian trees for varying coefficients | Many studies have reported associations between later-life cognition and
socioeconomic position in childhood, young adulthood, and mid-life. However,
the vast majority of these studies are unable to quantify how these
associations vary over time and with respect to several demographic factors.
Varying coefficient (VC) models, which treat the covariate effects in a linear
model as nonparametric functions of additional effect modifiers, offer an
appealing way to overcome these limitations. Unfortunately, state-of-the-art VC
modeling methods require computationally prohibitive parameter tuning or make
restrictive assumptions about the functional form of the covariate effects.
In response, we propose VCBART, which estimates the covariate effects in a VC
model using Bayesian Additive Regression Trees. With simple default
hyperparameter settings, VCBART outperforms existing methods in terms of
covariate effect estimation and prediction. Using VCBART, we predict the
cognitive trajectories of 4,167 subjects from the Health and Retirement Study
using multiple measures of socioeconomic position and physical health. We find
that socioeconomic position in childhood and young adulthood have small effects
that do not vary with age. In contrast, the effects of measures of mid-life
physical health tend to vary with respect to age, race, and marital status. An
R package implementing VC-BART is available at
https://github.com/skdeshpande91/VCBART
| stat |
Effects of Aggregation Methodology on Uncertain Spatiotemporal Data | Large spatiotemporal demand datasets can prove intractable for location
optimization problems, motivating the need to aggregate such data. However,
demand aggregation introduces error which impacts the results of the location
study. We introduce and apply a framework for comparing both deterministic and
stochastic aggregation methods using distance-based and volume-based
aggregation error metrics. In addition we introduce and apply weighted versions
of these metrics to account for the reality that demand events are
non-homogeneous. These metrics are applied to a large, highly variable,
spatiotemporal demand dataset of search and rescue events in the Pacific ocean.
Comparisons with these metrics between six quadrat aggregations of varying
scales and two zonal distribution models using hierarchical clustering is
conducted. We show that as quadrat fidelity increases the distance-based
aggregation error decreases, while the two deliberate zonal approaches further
reduce this error while utilizing fewer zones. However, the higher fidelity
aggregations have a detrimental effect on volume error. In addition, by
splitting the search and rescue dataset into a training and test set we show
that stochastic aggregation of this highly variable spatiotemporal demand
appears to be effective at simulating actual future demands.
| stat |
Fundamental Issues Regarding Uncertainties in Artificial Neural Networks | Artificial Neural Networks (ANNs) implement a specific form of multi-variate
extrapolation and will generate an output for any input pattern, even when
there is no similar training pattern. Extrapolations are not necessarily to be
trusted, and in order to support safety critical systems, we require such
systems to give an indication of the training sample related uncertainty
associated with their output. Some readers may think that this is a well known
issue which is already covered by the basic principles of pattern recognition.
We will explain below how this is not the case and how the conventional
(Likelihood estimate of) conditional probability of classification does not
correctly assess this uncertainty. We provide a discussion of the standard
interpretations of this problem and show how a quantitative approach based upon
long standing methods can be practically applied. The methods are illustrated
on the task of early diagnosis of dementing diseases using Magnetic Resonance
Imaging.
| stat |
OPENMENDEL: A Cooperative Programming Project for Statistical Genetics | Statistical methods for genomewide association studies (GWAS) continue to
improve. However, the increasing volume and variety of genetic and genomic data
make computational speed and ease of data manipulation mandatory in future
software. In our view, a collaborative effort of statistical geneticists is
required to develop open source software targeted to genetic epidemiology. Our
attempt to meet this need is called the OPENMENDELproject
(https://openmendel.github.io). It aims to (1) enable interactive and
reproducible analyses with informative intermediate results, (2) scale to big
data analytics, (3) embrace parallel and distributed computing, (4) adapt to
rapid hardware evolution, (5) allow cloud computing, (6) allow integration of
varied genetic data types, and (7) foster easy communication between
clinicians, geneticists, statisticians, and computer scientists. This article
reviews and makes recommendations to the genetic epidemiology community in the
context of the OPENMENDEL project.
| stat |
Flexible Bivariate INGARCH Process With a Broad Range of Contemporaneous
Correlation | We propose a novel flexible bivariate conditional Poisson (BCP)
INteger-valued Generalized AutoRegressive Conditional Heteroscedastic (INGARCH)
model for correlated count time series data. Our proposed BCP-INGARCH model is
mathematically tractable and has as the main advantage over existing bivariate
INGARCH models its ability to capture a broad range (both negative and
positive) of contemporaneous cross-correlation which is a non-trivial
advancement. Properties of stationarity and ergodicity for the BCP-INGARCH
process are developed. Estimation of the parameters is performed through
conditional maximum likelihood (CML) and finite sample behavior of the
estimators are investigated through simulation studies. Asymptotic properties
of the CML estimators are derived. Additional simulation studies compare and
contrast methods of obtaining standard errors of the parameter estimates, where
a bootstrap option is demonstrated to be advantageous. Hypothesis testing
methods for the presence of contemporaneous correlation between the time series
are presented and evaluated. We apply our methodology to monthly counts of
hepatitis cases at two nearby Brazilian cities, which are highly
cross-correlated. The data analysis demonstrates the importance of considering
a bivariate model allowing for a wide range of contemporaneous correlation in
real-life applications.
| stat |
On continual single index learning | In this paper, we generalize the problem of single index model to the context
of continual learning in which a learner is challenged with a sequence of tasks
one by one and the dataset of each task is revealed in an online fashion. We
propose a strategy that is able to learn a common single index for all tasks
and a specific link function for each task. The common single index allows to
transfer the informaton gained from the previous tasks to a new one. We provide
a theoretical analysis of our proposed strategy by proving some regret bounds.
Moreover, as a by-product from our work to provide an example of a within-task
algorithm, we develop a novel online algorithm for learning single index model
in an online setting and provide its regret bound.
| stat |
Reconstruction of stochastic 3-D signals with symmetric statistics from
2-D projection images motivated by cryo-electron microscopy | Cryo-electron microscopy provides 2-D projection images of the 3-D electron
scattering intensity of many instances of the particle under study (e.g., a
virus). Both symmetry (rotational point groups) and heterogeneity are important
aspects of biological particles and both aspects can be combined by describing
the electron scattering intensity of the particle as a stochastic process with
a symmetric probability law and therefore symmetric moments. A maximum
likelihood estimator implemented by an expectation-maximization algorithm is
described which estimates the unknown statistics of the electron scattering
intensity stochastic process from images of instances of the particle. The
algorithm is demonstrated on the bacteriophage HK97 and the virus N$\omega$V.
The results are contrasted with existing algorithms which assume that each
instance of the particle has the symmetry rather than the less restrictive
assumption that the probability law has the symmetry.
| stat |
Kernelized Complete Conditional Stein Discrepancy | Much of machine learning relies on comparing distributions with discrepancy
measures. Stein's method creates discrepancy measures between two distributions
that require only the unnormalized density of one and samples from the other.
Stein discrepancies can be combined with kernels to define kernelized Stein
discrepancies (KSDs). While kernels make Stein discrepancies tractable, they
pose several challenges in high dimensions. We introduce kernelized complete
conditional Stein discrepancies (KCC-SDs). Complete conditionals turn a
multivariate distribution into multiple univariate distributions. We show that
KCC-SDs distinguish distributions. To show the efficacy of KCC-SDs in
distinguishing distributions, we introduce a goodness-of-fit test using
KCC-SDs. We empirically show that KCC-SDs have higher power over baselines and
use KCC-SDs to assess sample quality in Markov chain Monte Carlo.
| stat |
When Do Neural Networks Outperform Kernel Methods? | For a certain scaling of the initialization of stochastic gradient descent
(SGD), wide neural networks (NN) have been shown to be well approximated by
reproducing kernel Hilbert space (RKHS) methods. Recent empirical work showed
that, for some classification tasks, RKHS methods can replace NNs without a
large loss in performance. On the other hand, two-layers NNs are known to
encode richer smoothness classes than RKHS and we know of special examples for
which SGD-trained NN provably outperform RKHS. This is true even in the wide
network limit, for a different scaling of the initialization.
How can we reconcile the above claims? For which tasks do NNs outperform
RKHS? If feature vectors are nearly isotropic, RKHS methods suffer from the
curse of dimensionality, while NNs can overcome it by learning the best
low-dimensional representation. Here we show that this curse of dimensionality
becomes milder if the feature vectors display the same low-dimensional
structure as the target function, and we precisely characterize this tradeoff.
Building on these results, we present a model that can capture in a unified
framework both behaviors observed in earlier work.
We hypothesize that such a latent low-dimensional structure is present in
image classification. We test numerically this hypothesis by showing that
specific perturbations of the training distribution degrade the performances of
RKHS methods much more significantly than NNs.
| stat |
Cosine Series Representation | This short paper is based on Chung et al. (2010), where the cosine series
representation (CSR) is used in modeling the shape of white matter fiber tracts
in diffusion tensor imaging(DTI) and Wang et al. (2018), where the method is
used to denoise EEG. The proposed explicit analytic approach offers far
superior flexibility in statistical modeling compared to the usual implicit
Fourier transform methods such as the discrete cosine transforms often used in
signal processing. The MATLAB codes and sample data can be obtained from
http://brainimaging.waisman.wisc.edu/~chung/tracts.
| stat |
Bayesian hierarchical space-time models to improve multispecies
assessment by combining observations from disparate fish surveys | Many wild species affected by human activities require multiple surveys with
differing designs to capture behavioural response to wide ranging habitat
conditions and map and quantify them. While data from for example intersecting
but disparate fish surveys using different gear, are widely available,
differences in design and methodology often limit their integration. Novel
statistical approaches which can draw on observations from diverse sources
could enhance our understanding of multiple species distributions
simultaneously and thus provide vital evidence needed to conserve their
populations and biodiversity at large. Using a novel Bayesian hierarchical
binomial-lognormal hurdle modelling approach within the INLA-SPDE framework, we
combined and analysed acoustic and bottom trawl survey data for herring, sprat
and northeast Atlantic mackerel in the North Sea. These models were implemented
using INLA-SPDE techniques. By accounting for gear-specific efficiencies across
surveys in addition to increased spatial coverage, we gained larger statistical
power with greatly minimised uncertainties in estimation. Our statistical
approach provides a methodological development to improve the evidence base for
multispecies assessment and marine ecosystem-based management. And on a broader
scale, it could be readily applied where disparate biological surveys and
sampling methods intersect, e.g. to provide information on biodiversity
patterns using global datasets of species distributions.
| stat |
Minimal Dispersion Approximately Balancing Weights: Asymptotic
Properties and Practical Considerations | Weighting methods are widely used to adjust for covariates in observational
studies, sample surveys, and regression settings. In this paper, we study a
class of recently proposed weighting methods which find the weights of minimum
dispersion that approximately balance the covariates. We call these weights
"minimal weights" and study them under a common optimization framework. The key
observation is the connection between approximate covariate balance and
shrinkage estimation of the propensity score. This connection leads to both
theoretical and practical developments. From a theoretical standpoint, we
characterize the asymptotic properties of minimal weights and show that, under
standard smoothness conditions on the propensity score function, minimal
weights are consistent estimates of the true inverse probability weights. Also,
we show that the resulting weighting estimator is consistent, asymptotically
normal, and semiparametrically efficient. From a practical standpoint, we
present a finite sample oracle inequality that bounds the loss incurred by
balancing more functions of the covariates than strictly needed. This
inequality shows that minimal weights implicitly bound the number of active
covariate balance constraints. We finally provide a tuning algorithm for
choosing the degree of approximate balance in minimal weights. We conclude the
paper with four empirical studies that suggest approximate balance is
preferable to exact balance, especially when there is limited overlap in
covariate distributions. In these studies, we show that the root mean squared
error of the weighting estimator can be reduced by as much as a half with
approximate balance.
| stat |
Joint, Partially-joint, and Individual Independent Component Analysis in
Multi-Subject fMRI Data | Objective: Joint analysis of multi-subject brain imaging datasets has wide
applications in biomedical engineering. In these datasets, some sources belong
to all subjects (joint), a subset of subjects (partially-joint), or a single
subject (individual). In this paper, this source model is referred to as
joint/partially-joint/individual multiple datasets multidimensional (JpJI-MDM),
and accordingly, a source extraction method is developed. Method: We present a
deflation-based algorithm utilizing higher order cumulants to analyze the
JpJI-MDM source model. The algorithm maximizes a cost function which leads to
an eigenvalue problem solved with thin-SVD (singular value decomposition)
factorization. Furthermore, we introduce the JpJI-feature which indicates the
spatial shape of each source and the amount of its jointness with other
subjects. We use this feature to determine the type of sources. Results: We
evaluate our algorithm by analyzing simulated data and two real functional
magnetic resonance imaging (fMRI) datasets. In our simulation study, we will
show that the proposed algorithm determines the type of sources with the
accuracy of 95% and 100% for 2-class and 3-class clustering scenarios,
respectively. Furthermore, our algorithm extracts meaningful joint and
partially-joint sources from the two real datasets, which are consistent with
the existing neuroscience studies. Conclusion: Our results in analyzing the
real datasets reveal that both datasets follow the JpJI-MDM source model. This
source model improves the accuracy of source extraction methods developed for
multi-subject datasets. Significance: The proposed joint blind source
separation algorithm is robust and avoids parameters which are difficult to
fine-tune.
| stat |
Shrinking the eigenvalues of M-estimators of covariance matrix | A highly popular regularized (shrinkage) covariance matrix estimator is the
shrinkage sample covariance matrix (SCM) which shares the same set of
eigenvectors as the SCM but shrinks its eigenvalues toward the grand mean of
the eigenvalues of the SCM. In this paper, a more general approach is
considered in which the SCM is replaced by an M-estimator of scatter matrix and
a fully automatic data adaptive method to compute the optimal shrinkage
parameter with minimum mean squared error is proposed. Our approach permits the
use of any weight function such as Gaussian, Huber's, Tyler's, or t-weight
functions, all of which are commonly used in M-estimation framework. Our
simulation examples illustrate that shrinkage M-estimators based on the
proposed optimal tuning combined with robust weight function do not loose in
performance to shrinkage SCM estimator when the data is Gaussian, but provide
significantly improved performance when the data is sampled from an unspecified
heavy-tailed elliptically symmetric distribution. Also, real-world and
synthetic stock market data validate the performance of the proposed method in
practical applications.
| stat |
Prediction-Based Decisions and Fairness: A Catalogue of Choices,
Assumptions, and Definitions | A recent flurry of research activity has attempted to quantitatively define
"fairness" for decisions based on statistical and machine learning (ML)
predictions. The rapid growth of this new field has led to wildly inconsistent
terminology and notation, presenting a serious challenge for cataloguing and
comparing definitions. This paper attempts to bring much-needed order.
First, we explicate the various choices and assumptions made---often
implicitly---to justify the use of prediction-based decisions. Next, we show
how such choices and assumptions can raise concerns about fairness and we
present a notationally consistent catalogue of fairness definitions from the ML
literature. In doing so, we offer a concise reference for thinking through the
choices, assumptions, and fairness considerations of prediction-based decision
systems.
| stat |
Decision tool and Sample Size Calculator for Composite Endpoints | Summary points:
- This article considers the combination of two binary or two time-to-event
endpoints to form the primary composite endpoint for leading a trial.
- It discusses the relative efficiency of choosing a composite endpoint over
one of its components in terms of: the frequencies of observing each component;
the relative treatment effect of the tested therapy; and the association
between both components.
- We highlight the very important role of the association between components
in choosing the most efficient endpoint to use as primary.
- For better grounded future trials, we recommend trialists to always
reporting the association between components of the composite endpoint.
- Common fallacies to note when using composite endpoints: i) composite
endpoints always imply higher power; ii) treatment effect on the composite
endpoint is similar to the average effects of its components; and iii) the
probability of observing the primary endpoint increases significantly.
| stat |
Dynamical systems theory for causal inference with application to
synthetic control methods | In this paper, we adopt results in nonlinear time series analysis for causal
inference in dynamical settings.~Our motivation is policy analysis with panel
data, particularly through the use of "synthetic control" methods. These
methods regress pre-intervention outcomes of the treated unit to outcomes from
a pool of control units, and then use the fitted regression model to estimate
causal effects post-intervention. In this setting, we propose to screen out
control units that have a weak dynamical relationship to the treated unit. In
simulations, we show that this method can mitigate bias from "cherry-picking"
of control units, which is usually an important concern. We illustrate on
real-world applications, including the tobacco legislation example of
\citet{Abadie2010}, and Brexit.
| stat |
Modelling Career Trajectories of Cricket Players Using Gaussian
Processes | In the sport of cricket, variations in a player's batting ability can usually
be measured on one of two scales. Short-term changes in ability that are
observed during a single innings, and long-term changes that are witnessed
between matches, over entire playing careers. To measure long-term variations,
we derive a Bayesian parametric model that uses a Gaussian process to measure
and predict how the batting abilities of international cricketers fluctuate
between innings. The model is fitted using nested sampling given its high
dimensionality and for ease of model comparison. Generally speaking, the
results support an anecdotal description of a typical sporting career. Young
players tend to begin their careers with some raw ability, which improves over
time as a result of coaching, experience and other external circumstances.
Eventually, players reach the peak of their career, after which ability tends
to decline. The model provides more accurate quantifications of current and
future player batting abilities than traditional cricketing statistics, such as
the batting average. The results allow us to identify which players are
improving or deteriorating in terms of batting ability, which has practical
implications in terms of player comparison, talent identification and team
selection policy.
| stat |
Bayesian Hierarchical Bernoulli-Weibull Mixture Model for Extremely Rare
Events | Estimating the duration of user behavior is a central concern for most
internet companies. Survival analysis is a promising method for analyzing the
expected duration of events and usually assumes the same survival function for
all subjects and the event will occur in the long run. However, such
assumptions are inappropriate when the users behave differently or some events
never occur for some users, i.e., the conversion period on web services of the
light users with no intention of behaving actively on the service. Especially,
if the proportion of inactive users is high, this assumption can lead to
undesirable results. To address these challenges, this paper proposes a mixture
model that separately addresses active and inactive individuals with a latent
variable. First, we define this specific problem setting and show the
limitations of conventional survival analysis in addressing this problem. We
demonstrate how naturally our Bernoulli-Weibull model can accommodate the
challenge. The proposed model was extended further to a Bayesian hierarchical
model to incorporate each subject's parameter, offering substantial
improvements over conventional, non-hierarchical models in terms of WAIC and
WBIC. Second, an experiment and extensive analysis were conducted using
real-world data from the Japanese job search website, CareerTrek, offered by
BizReach, Inc. In the analysis, some research questions are raised, such as the
difference in activation rate and conversion rate between user categories, and
how instantaneously the rate of event occurrence changes as time passes.
Quantitative answers and interpretations are assigned to them. Furthermore, the
model is inferred in a Bayesian manner, which enables us to represent the
uncertainty with a credible interval of the parameters and predictive
quantities.
| stat |
Pairwise likelihood estimation of latent autoregressive count models | Latent autoregressive models are useful time series models for the analysis
of infectious disease data. Evaluation of the likelihood function of latent
autoregressive models is intractable and its approximation through
simulation-based methods appears as a standard practice. Although simulation
methods may make the inferential problem feasible, they are often
computationally intensive and the quality of the numerical approximation may be
difficult to assess. We consider instead a weighted pairwise likelihood
approach and explore several computational and methodological aspects including
estimation of robust standard errors and the role of numerical integration. The
suggested approach is illustrated using monthly data on invasive meningococcal
disease infection in Greece and Italy.
| stat |
A Bayesian Functional Data Model for Surveys Collected under Informative
Sampling with Application to Mortality Estimation using NHANES | Functional data are often extremely high-dimensional and exhibit strong
dependence structures but can often prove valuable for both prediction and
inference. The literature on functional data analysis is well developed;
however, there has been very little work involving functional data in complex
survey settings. Motivated by physical activity monitor data from the National
Health and Nutrition Examination Survey (NHANES), we develop a Bayesian model
for functional covariates that can properly account for the survey design. Our
approach is intended for non-Gaussian data and can be applied in multivariate
settings. In addition, we make use of a variety of Bayesian modeling techniques
to ensure that the model is fit in a computationally efficient manner. We
illustrate the value of our approach through an empirical simulation study as
well as an example of mortality estimation using NHANES data.
| stat |
Multi-output Bus Travel Time Prediction with Convolutional LSTM Neural
Network | Accurate and reliable travel time predictions in public transport networks
are essential for delivering an attractive service that is able to compete with
other modes of transport in urban areas. The traditional application of this
information, where arrival and departure predictions are displayed on digital
boards, is highly visible in the city landscape of most modern metropolises.
More recently, the same information has become critical as input for
smart-phone trip planners in order to alert passengers about unreachable
connections, alternative route choices and prolonged travel times. More
sophisticated Intelligent Transport Systems (ITS) include the predictions of
connection assurance, i.e. to hold back services in case a connecting service
is delayed. In order to operate such systems, and to ensure the confidence of
passengers in the systems, the information provided must be accurate and
reliable. Traditional methods have trouble with this as congestion, and thus
travel time variability, increases in cities, consequently making travel time
predictions in urban areas a non-trivial task. This paper presents a system for
bus travel time prediction that leverages the non-static spatio-temporal
correlations present in urban bus networks, allowing the discovery of complex
patterns not captured by traditional methods. The underlying model is a
multi-output, multi-time-step, deep neural network that uses a combination of
convolutional and long short-term memory (LSTM) layers. The method is
empirically evaluated and compared to other popular approaches for link travel
time prediction and currently available services, including the currently
deployed model in Copenhagen, Denmark. We find that the proposed model
significantly outperforms all the other methods we compare with, and is able to
detect small irregular peaks in bus travel times very quickly.
| stat |
Hierarchical Multimodel Ensemble Estimates of Soil Water Retention with
Global Coverage | A correct quantification of mass and energy exchange processes among land
surface and atmosphere requires an accurate description of unsaturated soil
hydraulic properties. Soil pedotransfer functions (PTFs) have been widely used
to predict soil hydraulic parameters. Here, 13 PTFs were grouped according to
input data requirements and evaluated against a well-documented soil database
with global coverage. Weighted ensembles (calibrated by four groups and the
full 13-member set of PTFs) were shown to have improved performance over
individual PTFs in terms of root mean square error and other model selection
criteria. Global maps of soil water retention data from the ensemble models as
well as their uncertainty were provided. These maps demonstrate that five PTF
ensembles tend to have different estimates, especially in middle and high
latitudes in the Northern Hemisphere. Our full 13-member ensemble model
provides more accurate estimates than PTFs that are currently being used in
earth system models.
| stat |
Bayesian Uncertainty Estimation Under Complex Sampling | Social and economic studies are often implemented as complex survey designs.
For example, multistage, unequal probability sampling designs utilized by
federal statistical agencies are typically constructed to maximize the
efficiency of the target domain level estimator (e.g., indexed by geographic
area) within cost constraints for survey administration. Such designs may
induce dependence between the sampled units; for example, with employment of a
sampling step that selects geographically-indexed clusters of units. A
sampling-weighted pseudo-posterior distribution may be used to estimate the
population model on the observed sample. The dependence induced between
co-clustered units inflates the scale of the resulting pseudo-posterior
covariance matrix that has been shown to induce under coverage of the
credibility sets. By bridging results across Bayesian model mispecification and
survey sampling, we demonstrate that the scale and shape of the asymptotic
distributions are different between each of the pseudo-MLE, the
pseudo-posterior and the MLE under simple random sampling. Through insights
from survey sampling variance estimation and recent advances in computational
methods, we devise a correction applied as a simple and fast post-processing
step to MCMC draws of the pseudo-posterior distribution. This adjustment
projects the pseudo-posterior covariance matrix such that the nominal coverage
is approximately achieved. We make an application to the National Survey on
Drug Use and Health as a motivating example and we demonstrate the efficacy of
our scale and shape projection procedure on synthetic data on several common
archetypes of survey designs.
| stat |
Estimating Average Treatment Effects with Support Vector Machines | Support vector machine (SVM) is one of the most popular classification
algorithms in the machine learning literature. We demonstrate that SVM can be
used to balance covariates and estimate average causal effects under the
unconfoundedness assumption. Specifically, we adapt the SVM classifier as a
kernel-based weighting procedure that minimizes the maximum mean discrepancy
between the treatment and control groups while simultaneously maximizing
effective sample size. We also show that SVM is a continuous relaxation of the
quadratic integer program for computing the largest balanced subset,
establishing its direct relation to the cardinality matching method. Another
important feature of SVM is that the regularization parameter controls the
trade-off between covariate balance and effective sample size. As a result, the
existing SVM path algorithm can be used to compute the balance-sample size
frontier. We characterize the bias of causal effect estimation arising from
this trade-off, connecting the proposed SVM procedure to the existing kernel
balancing methods. Finally, we conduct simulation and empirical studies to
evaluate the performance of the proposed methodology and find that SVM is
competitive with the state-of-the-art covariate balancing methods.
| stat |
HIV-prevalence mapping using Small Area Estimation in Kenya, Tanzania,
and Mozambique at the first sub-national level | Local estimates of HIV-prevalence provide information that can be used to
target interventions and consequently increase the efficiency of the resources.
This closer-to-optimal allocation can lead to better health outcomes, including
the control of the disease spread, and for more people. Producing reliable
estimates at smaller geographical levels can be challenging and careful
consideration of the nature of the data and the epidemiologic rational is
needed. In this paper, we use the DHS data phase V to estimate HIV prevalence
at the first-subnational level in Kenya, Tanzania, and Mozambique. We fit the
data to a spatial random effect intrinsic conditional autoregressive (ICAR)
model to smooth the outcome. We also use a sampling specification from a
multistage cluster design. We found that Nyanza (P=14.2%) and Nairobi (P=7.8%)
in Kenya, Iringa (P=16.2%) and Dar es Salaam (P=10.1%) in Tanzania, and Gaza
(P=13.7%) and Maputo City (P=12.7%) in Mozambique are the regions with the
highest prevalence of HIV, within country. Our results are based on
statistically rigorous methods that allowed us to obtain an accurate visual
representation of the HIV prevalence in the subset of African countries we
chose. These results can help in identification and targeting of high-prevalent
regions to increase the supply of healthcare services to reduce the spread of
the disease and increase the health quality of people living with HIV.
| stat |
Multivariate outlier detection based on a robust Mahalanobis distance
with shrinkage estimators | A collection of robust Mahalanobis distances for multivariate outlier
detection is proposed, based on the notion of shrinkage. Robust intensity and
scaling factors are optimally estimated to define the shrinkage. Some
properties are investigated, such as affine equivariance and breakdown value.
The performance of the proposal is illustrated through the comparison to other
techniques from the literature, in a simulation study and with a real dataset.
The behavior when the underlying distribution is heavy-tailed or skewed, shows
the appropriateness of the method when we deviate from the common assumption of
normality. The resulting high correct detection rates and low false detection
rates in the vast majority of cases, as well as the significantly smaller
computation time shows the advantages of our proposal.
| stat |
Monitoring dynamic networks: a simulation-based strategy for comparing
monitoring methods and a comparative study | Recently there has been a lot of interest in monitoring and identifying
changes in dynamic networks, which has led to the development of a variety of
monitoring methods. Unfortunately, these methods have not been systematically
compared; moreover, new methods are often designed for a specialized use case.
In light of this, we propose the use of simulation to compare the performance
of network monitoring methods over a variety of dynamic network changes. Using
our family of simulated dynamic networks, we compare the performance of several
state-of-the-art social network monitoring methods in the literature. We
compare their performance over a variety of types of change; we consider both
increases in communication levels, node propensity change as well as changes in
community structure. We show that there does not exist one method that is
uniformly superior to the others; the best method depends on the context and
the type of change one wishes to detect. As such, we conclude that a variety of
methods is needed for network monitoring and that it is important to understand
in which scenarios a given method is appropriate.
| stat |
Cohort State-Transition Models in R: A Tutorial | Decision models can synthesize evidence from different sources to simulate
long-term consequences of different strategies in the presence of uncertainty.
Cohort state-transition models (cSTM) are decision models commonly used in
medical decision making to simulate hypothetical cohorts' transitions across
various health states over time. This tutorial shows how to implement cSTMs in
R, an open-source mathematical and statistical programming language. As an
example, we use a previously published cSTM-based cost-effectiveness analysis.
With this example, we illustrate both time-independent cSTMs, where transition
probabilities are constant over time, and time-dependent cSTMs, where
transition probabilities vary by age and are dependent on time spent in a
health state (state residence). We also illustrate how to compute various
epidemiological outcomes of interest, such as survival and prevalence. We
demonstrate how to calculate economic outcomes and conducting a
cost-effectiveness analysis of multiple strategies using the example model, and
provide additional resources to conduct probabilistic sensitivity analyses. We
provide a link to a public repository with all the R code described in this
tutorial that can be used to replicate the example or be adapted for various
decision modeling applications.
| stat |
Distilling with Performance Enhanced Students | The task of accelerating large neural networks on general purpose hardware
has, in recent years, prompted the use of channel pruning to reduce network
size. However, the efficacy of pruning based approaches has since been called
into question. In this paper, we turn to distillation for model
compression---specifically, attention transfer---and develop a simple method
for discovering performance enhanced student networks. We combine channel
saliency metrics with empirical observations of runtime performance to design
more accurate networks for a given latency budget. We apply our methodology to
residual and densely-connected networks, and show that we are able to find
resource-efficient student networks on different hardware platforms while
maintaining very high accuracy. These performance-enhanced student networks
achieve up to 10% boosts in top-1 ImageNet accuracy over their channel-pruned
counterparts for the same inference time.
| stat |
Quantifying the Uncertainty in Model Parameters Using Gaussian
Process-Based Markov Chain Monte Carlo: An Application to Cardiac
Electrophysiological Models | Estimation of patient-specific model parameters is important for personalized
modeling, although sparse and noisy clinical data can introduce significant
uncertainty in the estimated parameter values. This importance source of
uncertainty, if left unquantified, will lead to unknown variability in model
outputs that hinder their reliable adoptions. Probabilistic estimation model
parameters, however, remains an unresolved challenge because standard Markov
Chain Monte Carlo sampling requires repeated model simulations that are
computationally infeasible. A common solution is to replace the simulation
model with a computationally-efficient surrogate for a faster sampling.
However, by sampling from an approximation of the exact posterior probability
density function (pdf) of the parameters, the efficiency is gained at the
expense of sampling accuracy. In this paper, we address this issue by
integrating surrogate modeling into Metropolis Hasting (MH) sampling of the
exact posterior pdfs to improve its acceptance rate. It is done by first
quickly constructing a Gaussian process (GP) surrogate of the exact posterior
pdfs using deterministic optimization. This efficient surrogate is then used to
modify commonly-used proposal distributions in MH sampling such that only
proposals accepted by the surrogate will be tested by the exact posterior pdf
for acceptance/rejection, reducing unnecessary model simulations at unlikely
candidates. Synthetic and real-data experiments using the presented method show
a significant gain in computational efficiency without compromising the
accuracy. In addition, insights into the non-identifiability and heterogeneity
of tissue properties can be gained from the obtained posterior distributions.
| stat |
Interpretable Stein Goodness-of-fit Tests on Riemannian Manifolds | In many applications, we encounter data on Riemannian manifolds such as torus
and rotation groups. Standard statistical procedures for multivariate data are
not applicable to such data. In this study, we develop goodness-of-fit testing
and interpretable model criticism methods for general distributions on
Riemannian manifolds, including those with an intractable normalization
constant. The proposed methods are based on extensions of kernel Stein
discrepancy, which are derived from Stein operators on Riemannian manifolds. We
discuss the connections between the proposed tests with existing ones and
provide a theoretical analysis of their asymptotic Bahadur efficiency.
Simulation results and real data applications show the validity of the proposed
methods.
| stat |
B-CONCORD -- A scalable Bayesian high-dimensional precision matrix
estimation procedure | Sparse estimation of the precision matrix under high-dimensional scaling
constitutes a canonical problem in statistics and machine learning. Numerous
regression and likelihood based approaches, many frequentist and some Bayesian
in nature have been developed. Bayesian methods provide direct uncertainty
quantification of the model parameters through the posterior distribution and
thus do not require a second round of computations for obtaining debiased
estimates of the model parameters and their confidence intervals. However, they
are computationally expensive for settings involving more than 500 variables.
To that end, we develop B-CONCORD for the problem at hand, a Bayesian analogue
of the CONvex CORrelation selection methoD (CONCORD) introduced by Khare et al.
(2015). B-CONCORD leverages the CONCORD generalized likelihood function
together with a spike-and-slab prior distribution to induce sparsity in the
precision matrix parameters. We establish model selection and estimation
consistency under high-dimensional scaling; further, we develop a procedure
that refits only the non-zero parameters of the precision matrix, leading to
significant improvements in the estimates in finite samples. Extensive
numerical work illustrates the computational scalability of the proposed
approach vis-a-vis competing Bayesian methods, as well as its accuracy.
| stat |
Empirical Bayes Matrix Factorization | Matrix factorization methods - including Factor analysis (FA), and Principal
Components Analysis (PCA) - are widely used for inferring and summarizing
structure in multivariate data. Many matrix factorization methods exist,
corresponding to different assumptions on the elements of the underlying matrix
factors. For example, many recent methods use a penalty or prior distribution
to achieve sparse representations ("Sparse FA/PCA"). Here we introduce a
general Empirical Bayes approach to matrix factorization (EBMF), whose key
feature is that it uses the observed data to estimate prior distributions on
matrix elements. We derive a correspondingly-general variational fitting
algorithm, which reduces fitting EBMF to solving a simpler problem - the
so-called "normal means" problem. We implement this general algorithm, but
focus particular attention on the use of sparsity-inducing priors that are
uni-modal at 0. This yields a sparse EBMF approach - essentially a version of
sparse FA/PCA - that automatically adapts the amount of sparsity to the data.
We demonstrate the benefits of our approach through both numerical comparisons
with competing methods and through analysis of data from the GTEx (Genotype
Tissue Expression) project on genetic associations across 44 human tissues. In
numerical comparisons EBMF often provides more accurate inferences than other
methods. In the GTEx data, EBMF identifies interpretable structure that
concords with known relationships among human tissues. Software implementing
our approach is available at https://github.com/stephenslab/flashr
| stat |
On the Connection Between Adversarial Robustness and Saliency Map
Interpretability | Recent studies on the adversarial vulnerability of neural networks have shown
that models trained to be more robust to adversarial attacks exhibit more
interpretable saliency maps than their non-robust counterparts. We aim to
quantify this behavior by considering the alignment between input image and
saliency map. We hypothesize that as the distance to the decision boundary
grows,so does the alignment. This connection is strictly true in the case of
linear models. We confirm these theoretical findings with experiments based on
models trained with a local Lipschitz regularization and identify where the
non-linear nature of neural networks weakens the relation.
| stat |
Statistical Analysis of Behavioral Intention Towards Private Umbilical
Cord Blood Banking | In this paper, we propose a conceptual framework to identify the key
dimensions affecting behavioral intention to bank umbilical cord blood in Iran.
We examine the impact of awareness, reference group, usability, disease
history, and price on perceived risk and behavioral intention to use umbilical
cord blood banking service. To evaluate the proposed model of umbilical cord
blood banking behavioral intention and to test our hypotheses, we apply field
exploratory research. We use a five-point Likert scale to form a questionnaire
to collect the data. The model is estimated with a sample of 242 Royan cord
blood bank customers in Tehran. We use Pearson correlation and structural
equation modeling to analyze the structural relationships between research
variables, perceived risk, and behavioral intention. This research gives
novelty on the determinants of behavioral intention in private umbilical cord
blood banking, which adds value to literature and future managerial practices.
Results show that usability is the primary determinant in cord blood banking.
| stat |
Error Correcting Algorithms for Sparsely Correlated Regressors | Autonomy and adaptation of machines requires that they be able to measure
their own errors. We consider the advantages and limitations of such an
approach when a machine has to measure the error in a regression task. How can
a machine measure the error of regression sub-components when it does not have
the ground truth for the correct predictions? A compressed sensing approach
applied to the error signal of the regressors can recover their precision error
without any ground truth. It allows for some regressors to be \emph{strongly
correlated} as long as not too many are so related. Its solutions, however, are
not unique - a property of ground truth inference solutions. Adding
$\ell_1$--minimization as a condition can recover the correct solution in
settings where error correction is possible. We briefly discuss the similarity
of the mathematics of ground truth inference for regressors to that for
classifiers.
| stat |
A distribution-free smoothed combination method of biomarkers to improve
diagnostic accuracy in multi-category classification | Results from multiple diagnostic tests are usually combined to improve the
overall diagnostic accuracy. For binary classification, maximization of the
empirical estimate of the area under the receiver operating characteristic
(ROC) curve is widely adopted to produce the optimal linear combination of
multiple biomarkers. In the presence of large number of biomarkers, this method
proves to be computationally expensive and difficult to implement since it
involves maximization of a discontinuous, non-smooth function for which
gradient-based methods cannot be used directly. Complexity of this problem
increases when the classification problem becomes multi-category. In this
article, we develop a linear combination method that maximizes a smooth
approximation of the empirical Hypervolume Under Manifolds (HUM) for
multi-category outcome. We approximate HUM by replacing the indicator function
with the sigmoid function or normal cumulative distribution function (CDF).
With the above smooth approximations, efficient gradient-based algorithms can
be employed to obtain better solution with less computing time. We show that
under some regularity conditions, the proposed method yields consistent
estimates of the coefficient parameters. We also derive the asymptotic
normality of the coefficient estimates. We conduct extensive simulations to
examine our methods. Under different simulation scenarios, the proposed methods
are compared with other existing methods and are shown to outperform them in
terms of diagnostic accuracy. The proposed method is illustrated using two real
medical data sets.
| stat |
Discrete distributions from a Markov chain | A discrete-time stochastic process derived from a model of basketball is used
to generalize any discrete distribution. The generalized distributions can have
one or two more parameters than the parent distribution. Those derived from
binomial, Poisson and negative binomial distributions can be underdispersed or
overdispersed. The mean can be simply expressed in terms of model parameters,
thus making inference for the mean straightforward. Probabilities can be
quickly computed, enabling likelihood-based inference. Random number generation
is also straightforward. The properties of some of the new distributions are
described and their use is illustrated with examples.
| stat |
Modeling Weather-induced Home Insurance Risks with Support Vector
Machine Regression | Insurance industry is one of the most vulnerable sectors to climate change.
Assessment of future number of claims and incurred losses is critical for
disaster preparedness and risk management. In this project, we study the effect
of precipitation on a joint dynamics of weather-induced home insurance claims
and losses. We discuss utility and limitations of such machine learning
procedures as Support Vector Machines and Artificial Neural Networks, in
forecasting future claim dynamics and evaluating associated uncertainties. We
illustrate our approach by application to attribution analysis and forecasting
of weather-induced home insurance claims in a middle-sized city in the Canadian
Prairies.
| stat |
Using Multiple Imputation to Classify Potential Outcomes Subgroups | With medical tests becoming increasingly available, concerns about
over-testing and over-treatment dramatically increase. Hence, it is important
to understand the influence of testing on treatment selection in general
practice. Most statistical methods focus on average effects of testing on
treatment decisions. However, this may be ill-advised, particularly for patient
subgroups that tend not to benefit from such tests. Furthermore, missing data
are common, representing large and often unaddressed threats to the validity of
statistical methods. Finally, it is desirable to conduct analyses that can be
interpreted causally. We propose to classify patients into four potential
outcomes subgroups, defined by whether or not a patient's treatment selection
is changed by the test result and by the direction of how the test result
changes treatment selection. This subgroup classification naturally captures
the differential influence of medical testing on treatment selections for
different patients, which can suggest targets to improve the utilization of
medical tests. We can then examine patient characteristics associated with
patient potential outcomes subgroup memberships. We used multiple imputation
methods to simultaneously impute the missing potential outcomes as well as
regular missing values. This approach can also provide estimates of many
traditional causal quantities. We find that explicitly incorporating causal
inference assumptions into the multiple imputation process can improve the
precision for some causal estimates of interest. We also find that bias can
occur when the potential outcomes conditional independence assumption is
violated; sensitivity analyses are proposed to assess the impact of this
violation. We applied the proposed methods to examine the influence of 21-gene
assay, the most commonly used genomic test, on chemotherapy selection among
breast cancer patients.
| stat |
Spatial methods and their applications to environmental and climate data | Environmental and climate processes are often distributed over large
space-time domains. Their complexity and the amount of available data make
modelling and analysis a challenging task. Statistical modelling of environment
and climate data can have several different motivations including
interpretation or characterisation of the data. Results from statistical
analysis are often used as a integral part of larger environmental studies.
Spatial statistics is an active and modern statistical field, concerned with
the quantitative analysis of spatial data; their dependencies and
uncertainties. Spatio-temporal statistics extends spatial statistics through
the addition of time to the, two or three, spatial dimensions. The focus of
this introductory paper is to provide an overview of spatial methods and their
application to environmental and climate data. This paper also gives an
overview of several important topics including large data sets and
non-stationary covariance structures. Further, it is discussed how Bayesian
hierarchical models can provide a flexible way of constructing models.
Hierarchical models may seem to be a good solution, but they have challenges of
their own such as, parameter estimation. Finally, the application of
spatio-temporal models to the LANDCLIM data (LAND cover - CLIMate interactions
in NW Europe during the Holocene) will be discussed.
| stat |
A Vecchia Approximation for High-Dimensional Gaussian Cumulative
Distribution Functions Arising from Spatial Data | We introduce an approach to quickly and accurately approximate the cumulative
distribution function of multivariate Gaussian distributions arising from
spatial Gaussian processes. This approximation is trivially parallelizable and
simple to implement using standard software. We demonstrate its accuracy and
computational efficiency in a series of simulation experiments and apply it to
analyzing the joint tail of a large precipitation dataset using a
recently-proposed scale mixture model for spatial extremes. This dataset is
many times larger than what was previously considered possible to fit using
preferred inferential techniques.
| stat |
Sparse Quantized Spectral Clustering | Given a large data matrix, sparsifying, quantizing, and/or performing other
entry-wise nonlinear operations can have numerous benefits, ranging from
speeding up iterative algorithms for core numerical linear algebra problems to
providing nonlinear filters to design state-of-the-art neural network models.
Here, we exploit tools from random matrix theory to make precise statements
about how the eigenspectrum of a matrix changes under such nonlinear
transformations. In particular, we show that very little change occurs in the
informative eigenstructure even under drastic sparsification/quantization, and
consequently that very little downstream performance loss occurs with very
aggressively sparsified or quantized spectral clustering. We illustrate how
these results depend on the nonlinearity, we characterize a phase transition
beyond which spectral clustering becomes possible, and we show when such
nonlinear transformations can introduce spurious non-informative eigenvectors.
| stat |
Screening Sinkhorn Algorithm for Regularized Optimal Transport | We introduce in this paper a novel strategy for efficiently approximating the
Sinkhorn distance between two discrete measures. After identifying neglectable
components of the dual solution of the regularized Sinkhorn problem, we propose
to screen those components by directly setting them at that value before
entering the Sinkhorn problem. This allows us to solve a smaller Sinkhorn
problem while ensuring approximation with provable guarantees. More formally,
the approach is based on a new formulation of dual of Sinkhorn divergence
problem and on the KKT optimality conditions of this problem, which enable
identification of dual components to be screened. This new analysis leads to
the Screenkhorn algorithm. We illustrate the efficiency of Screenkhorn on
complex tasks such as dimensionality reduction and domain adaptation involving
regularized optimal transport.
| stat |
Testing and estimation of clustered signals | We propose a change-point detection method for large scale multiple testing
problems with data having clustered signals. Unlike the classic change-point
setup, the signals can vary in size within a cluster. The clustering structure
on the signals enables us to effectively delineate the boundaries between
signal and non-signal segments. New test statistics are proposed for
observations from one and/or multiple realizations. Their asymptotic
distributions are derived. We also study the associated variance estimation
problem. We allow the variances to be heteroscedastic in the multiple
realization case, which substantially expands the applicability of the proposed
method. Simulation studies demonstrate that the proposed approach has a
favorable performance. Our procedure is applied to {an array based Comparative
Genomic Hybridization (aCGH)} dataset.
| stat |
Tractable structured natural gradient descent using local
parameterizations | Natural-gradient descent on structured parameter spaces (e.g., low-rank
covariances) is computationally challenging due to complicated inverse
Fisher-matrix computations. We address this issue for optimization, inference,
and search problems by using \emph{local-parameter coordinates}. Our method
generalizes an existing evolutionary-strategy method, recovers Newton and
Riemannian-gradient methods as special cases, and also yields new tractable
natural-gradient algorithms for learning flexible covariance structures of
Gaussian and Wishart-based distributions via \emph{matrix groups}. We show
results on a range of applications on deep learning, variational inference, and
evolution strategies. Our work opens a new direction for scalable structured
geometric methods via local parameterizations.
| stat |
Calculating the Expected Value of Sample Information in Practice:
Considerations from Three Case Studies | Investing efficiently in future research to improve policy decisions is an
important goal. Expected Value of Sample Information (EVSI) can be used to
select the specific design and sample size of a proposed study by assessing the
benefit of a range of different studies. Estimating EVSI with the standard
nested Monte Carlo algorithm has a notoriously high computational burden,
especially when using a complex decision model or when optimizing over study
sample sizes and designs. Therefore, a number of more efficient EVSI
approximation methods have been developed. However, these approximation methods
have not been compared and therefore their relative advantages and
disadvantages are not clear. A consortium of EVSI researchers, including the
developers of several approximation methods, compared four EVSI methods using
three previously published health economic models. The examples were chosen to
represent a range of real-world contexts, including situations with multiple
study outcomes, missing data, and data from an observational rather than a
randomized study. The computational speed and accuracy of each method were
compared, and the relative advantages and implementation challenges of the
methods were highlighted. In each example, the approximation methods took
minutes or hours to achieve reasonably accurate EVSI estimates, whereas the
traditional Monte Carlo method took weeks. Specific methods are particularly
suited to problems where we wish to compare multiple proposed sample sizes,
when the proposed sample size is large, or when the health economic model is
computationally expensive. All the evaluated methods gave estimates similar to
those given by traditional Monte Carlo, suggesting that EVSI can now be
efficiently computed with confidence in realistic examples.
| stat |
Non-reversible Markov chain Monte Carlo for sampling of districting maps | Evaluating the degree of partisan districting (Gerrymandering) in a
statistical framework typically requires an ensemble of districting plans which
are drawn from a prescribed probability distribution that adheres to a
realistic and non-partisan criteria. In this article we introduce novel
non-reversible Markov chain Monte-Carlo (MCMC) methods for the sampling of such
districting plans which have improved mixing properties in comparison to
previously used (reversible) MCMC algorithms. In doing so we extend the current
framework for construction of non-reversible Markov chains on discrete sampling
spaces by considering a generalization of skew detailed balance. We provide a
detailed description of the proposed algorithms and evaluate their performance
in numerical experiments.
| stat |
Calibration procedures for approximate Bayesian credible sets | We develop and apply two calibration procedures for checking the coverage of
approximate Bayesian credible sets including intervals estimated using Monte
Carlo methods. The user has an ideal prior and likelihood, but generates a
credible set for an approximate posterior which is not proportional to the
product of ideal likelihood and prior. We estimate the realised posterior
coverage achieved by the approximate credible set. This is the coverage of the
unknown ``true'' parameter if the data are a realisation of the user's ideal
observation model conditioned on the parameter, and the parameter is a draw
from the user's ideal prior. In one approach we estimate the posterior coverage
at the data by making a semi-parametric logistic regression of binary coverage
outcomes on simulated data against summary statistics evaluated on simulated
data. In another we use Importance Sampling from the approximate posterior,
windowing simulated data to fall close to the observed data. We illustrate our
methods on four examples.
| stat |
A two-level Kriging-based approach with active learning for solving
time-variant risk optimization problems | Several methods have been proposed in the literature to solve
reliability-based optimization problems, where failure probabilities are design
constraints. However, few methods address the problem of life-cycle cost or
risk optimization, where failure probabilities are part of the objective
function. Moreover, few papers in the literature address time-variant
reliability problems in life-cycle cost or risk optimization formulations; in
particular, because most often computationally expensive Monte Carlo simulation
is required. This paper proposes a numerical framework for solving general risk
optimization problems involving time-variant reliability analysis. To alleviate
the computational burden of Monte Carlo simulation, two adaptive coupled
surrogate models are used: the first one to approximate the objective function,
and the second one to approximate the quasi-static limit state function. An
iterative procedure is implemented for choosing additional support points to
increase the accuracy of the surrogate models. Three application problems are
used to illustrate the proposed approach. Two examples involve random load and
random resistance degradation processes. The third problem is related to
load-path dependent failures. This subject had not yet been addressed in the
context of risk-based optimization. It is shown herein that accurate solutions
are obtained, with extremely limited numbers of objective function and limit
state functions calls.
| stat |
Distributed Feature Screening via Componentwise Debiasing | Feature screening is a powerful tool in the analysis of high dimensional
data. When the sample size $N$ and the number of features $p$ are both large,
the implementation of classic screening methods can be numerically challenging.
In this paper, we propose a distributed screening framework for big data setup.
In the spirit of "divide-and-conquer", the proposed framework expresses a
correlation measure as a function of several component parameters, each of
which can be distributively estimated using a natural U-statistic from data
segments. With the component estimates aggregated, we obtain a final
correlation estimate that can be readily used for screening features. This
framework enables distributed storage and parallel computing and thus is
computationally attractive. Due to the unbiased distributive estimation of the
component parameters, the final aggregated estimate achieves a high accuracy
that is insensitive to the number of data segments $m$ specified by the problem
itself or to be chosen by users. Under mild conditions, we show that the
aggregated correlation estimator is as efficient as the classic centralized
estimator in terms of the probability convergence bound; the corresponding
screening procedure enjoys sure screening property for a wide range of
correlation measures. The promising performances of the new method are
supported by extensive numerical examples.
| stat |
Perturbed factor analysis: Accounting for group differences in exposure
profiles | In this article, we investigate group differences in phthalate exposure
profiles using NHANES data. Phthalates are a family of industrial chemicals
used in plastics and as solvents. There is increasing evidence of adverse
health effects of exposure to phthalates on reproduction and neuro-development,
and concern about racial disparities in exposure. We would like to identify a
single set of low-dimensional factors summarizing exposure to different
chemicals, while allowing differences across groups. Improving on current
multi-group additive factor models, we propose a class of Perturbed Factor
Analysis (PFA) models that assume a common factor structure after perturbing
the data via multiplication by a group-specific matrix. Bayesian inference
algorithms are defined using a matrix normal hierarchical model for the
perturbation matrices. The resulting model is just as flexible as current
approaches in allowing arbitrarily large differences across groups but has
substantial advantages that we illustrate in simulation studies. Applying PFA
to NHANES data, we learn common factors summarizing exposures to phthalates,
while showing clear differences across groups.
| stat |
Bayesian estimation of trend components within Markovian
regime-switching models for wholesale electricity prices: an application to
the South Australian wholesale electricity market | We discuss and extend methods for estimating Markovian-Regime-Switching (MRS)
and trend models for wholesale electricity prices. We argue the existing
methods of trend estimation used in the electricity price modelling literature
either require an ambiguous definition of an extreme price, or lead to issues
when implementing model selection [23]. The first main contribution of this
paper is to design and infer a model which has a model-based definition of
extreme prices and permits the use of model selection criteria. Due to the
complexity of the MRS models inference is not straightforward. In the existing
literature an approximate EM algorithm is used [26]. Another contribution of
this paper is to implement exact inference in a Bayesian setting. This also
allows the use of posterior predictive checks to assess model fit. We
demonstrate the methodologies with South Australian electricity market.
| stat |
Locally Linear Embedding and its Variants: Tutorial and Survey | This is a tutorial and survey paper for Locally Linear Embedding (LLE) and
its variants. The idea of LLE is fitting the local structure of manifold in the
embedding space. In this paper, we first cover LLE, kernel LLE, inverse LLE,
and feature fusion with LLE. Then, we cover out-of-sample embedding using
linear reconstruction, eigenfunctions, and kernel mapping. Incremental LLE is
explained for embedding streaming data. Landmark LLE methods using the Nystrom
approximation and locally linear landmarks are explained for big data
embedding. We introduce the methods for parameter selection of number of
neighbors using residual variance, Procrustes statistics, preservation
neighborhood error, and local neighborhood selection. Afterwards, Supervised
LLE (SLLE), enhanced SLLE, SLLE projection, probabilistic SLLE, supervised
guided LLE (using Hilbert-Schmidt independence criterion), and semi-supervised
LLE are explained for supervised and semi-supervised embedding. Robust LLE
methods using least squares problem and penalty functions are also introduced
for embedding in the presence of outliers and noise. Then, we introduce fusion
of LLE with other manifold learning methods including Isomap (i.e., ISOLLE),
principal component analysis, Fisher discriminant analysis, discriminant LLE,
and Isotop. Finally, we explain weighted LLE in which the distances,
reconstruction weights, or the embeddings are adjusted for better embedding; we
cover weighted LLE for deformed distributed data, weighted LLE using
probability of occurrence, SLLE by adjusting weights, modified LLE, and
iterative LLE.
| stat |
Latent group structure and regularized regression | Regression models generally assume that the conditional distribution of
response Y given features X is the same for all samples. For heterogeneous data
with distributional differences among latent groups, standard regression models
are ill-equipped, especially in large multivariate problems where hidden
heterogeneity can easily pass undetected. To allow for robust and interpretable
regression modeling in this setting we propose a class of regularized mixture
models that couples together both the multivariate distribution of X and the
conditional Y | X. This joint modeling approach offers a novel way to deal with
suspected distributional shifts, which allows for automatic control of
confounding by latent group structure and delivers scalable, sparse solutions.
Estimation is handled via an expectation-maximization algorithm, whose
convergence is established theoretically. We illustrate the key ideas via
empirical examples.
| stat |
Measurement Error in Meta-Analysis (MEMA) -- a Bayesian framework for
continuous outcome data | Ideally, a meta-analysis will summarize data from several unbiased studies.
Here we consider the less than ideal situation in which contributing studies
may be compromised by measurement error. Measurement error affects every study
design, from randomized controlled trials to retrospective observational
studies. We outline a flexible Bayesian framework for continuous outcome data
which allows one to obtain appropriate point and interval estimates with
varying degrees of prior knowledge about the magnitude of the measurement
error. We also demonstrate how, if individual-participant data (IPD) are
available, the Bayesian meta-analysis model can adjust for multiple
participant-level covariates, measured with or without measurement error.
| stat |
The Discrete Adjoint Method: Efficient Derivatives for Functions of
Discrete Sequences | Gradient-based techniques are becoming increasingly critical in quantitative
fields, notably in statistics and computer science. The utility of these
techniques, however, ultimately depends on how efficiently we can evaluate the
derivatives of the complex mathematical functions that arise in applications.
In this paper we introduce a discrete adjoint method that efficiently evaluates
derivatives for functions of discrete sequences.
| stat |
Multidimensional Scaling for Big Data | We present a set of algorithms for Multidimensional Scaling (MDS) to be used
with large datasets. MDS is a statistic tool for reduction of dimensionality,
using as input a distance matrix of dimensions $n \times n$. When $n$ is large,
classical algorithms suffer from computational problems and MDS configuration
can not be obtained. In this paper we address these problems by means of three
algorithms: Divide and Conquer MDS, Fast MDS and MDS based on Gower
interpolation (the first and the last being original proposals). The main ideas
of these methods are based on partitioning the dataset into small pieces, where
classical MDS methods can work. In order to check the performance of the
algorithms as well as to compare them, we do a simulation study. This study
points out that Fast MDS and MDS based on Gower interpolation are appropriated
to use when $n$ is large. Although Divide and Conquer MDS is not as fast as the
other two algorithms, it is the best method that captures the variance of the
original data.
| stat |
Highest CO$_2$ Emissions Scenarios Are Extreme Given Observations and
Expert Judgements | Probabilistic projections of baseline future carbon emissions are important
for sound climate risk management. Deep uncertainty surrounds many drivers of
projected emissions. For example, there is no consensus about estimates of
fossil fuel resources. We use a simple integrated assessment model to make
probabilistic projections of carbon emissions through 2100. We find that, given
current mitigation policies and in the absence of negative emissions
technologies, more moderate scenarios used by the Intergovernmental Panel on
Climate Change are more likely than the extreme high or low scenarios. This
finding is robust to a variety of assumptions about fossil fuel resource
constraints and decarbonization rates. However, the most likely range for
cumulative CO$_2$ emissions from 2018--2100 across our varying assumptions is
700--1800 GtC. Much more aggressive mitigation will be required to reliably
achieve the $2^\circ$C Paris Agreement target.
| stat |
Modeling the Dynamics of the COVID-19 Population in Australia: A
Probabilistic Analysis | The novel Corona Virus COVID-19 arrived on Australian shores around 25
January 2020. This paper presents a novel method of dynamically modeling and
forecasting the COVID-19 pandemic in Australia with a high degree of accuracy
and in a timely manner using limited data; a valuable resource that can be used
to guide government decision-making on societal restrictions on a daily and/or
weekly basis. The "partially-observable stochastic process" used in this study
predicts not only the future actual values with extremely low error, but also
the percentage of unobserved COVID-19 cases in the population. The model can
further assist policy makers to assess the effectiveness of several possible
alternative scenarios in their decision-making processes.
| stat |
Accuracy of the Epic Sepsis Prediction Model in a Regional Health System | Interest in an electronic health record-based computational model that can
accurately predict a patient's risk of sepsis at a given point in time has
grown rapidly in the last several years. Like other EHR vendors, the Epic
Systems Corporation has developed a proprietary sepsis prediction model (ESPM).
Epic developed the model using data from three health systems and penalized
logistic regression. Demographic, comorbidity, vital sign, laboratory,
medication, and procedural variables contribute to the model. The objective of
this project was to compare the predictive performance of the ESPM with a
regional health system's current Early Warning Score-based sepsis detection
program.
| stat |
Simultaneous Inference Under the Vacuous Orientation Assumption | I propose a novel approach to simultaneous inference that alleviates the need
to specify a correlational structure among marginal errors. The vacuous
orientation assumption retains what the normal i.i.d. assumption implies about
the distribution of error configuration, but relaxes the implication that the
error orientation is isotropic. When a large number of highly dependent
hypotheses are tested simultaneously, the proposed model produces calibrated
posterior inference by leveraging the logical relationship among them. This
stands in contrast to the conservative performance of the Bonferroni
correction, even if neither approaches makes assumptions about error
dependence. The proposed model employs the Dempster-Shafer Extended Calculus of
Probability, and delivers posterior inference in the form of stochastic
three-valued logic.
| stat |
Bayesian nonparametric panel Markov-switching GARCH models | This paper introduces a new model for panel data with Markov-switching GARCH
effects. The model incorporates a series-specific hidden Markov chain process
that drives the GARCH parameters. To cope with the high-dimensionality of the
parameter space, the paper exploits the cross-sectional clustering of the
series by first assuming a soft parameter pooling through a hierarchical prior
distribution with two-step procedure, and then introducing clustering effects
in the parameter space through a nonparametric prior distribution. The model
and the proposed inference are evaluated through a simulation experiment. The
results suggest that the inference is able to recover the true value of the
parameters and the number of groups in each regime. An empirical application to
78 assets of the SP\&100 index from $6^{th}$ January 2000 to $3^{rd}$ October
2020 is also carried out by using a two-regime Markov switching GARCH model.
The findings shows the presence of 2 and 3 clusters among the constituents in
the first and second regime, respectively.
| stat |
Defying the Circadian Rhythm: Clustering Participant Telemetry in the UK
Biobank Data | The UK Biobank dataset follows over 500,000 volunteers and contains a diverse
set of information related to societal outcomes. Among this vast collection, a
large quantity of telemetry collected from wrist-worn accelerometers provides a
snapshot of participant activity. Using this data, a population of shift
workers, subjected to disrupted circadian rhythms, is analysed using a mixture
model-based approach to yield protective effects from physical activity on
survival outcomes. In this paper, we develop a scalable, standardized, and
unique methodology that efficiently clusters a vast quantity of participant
telemetry. By building upon the work of Doherty et al. (2017), we introduce a
standardized, low-dimensional feature for clustering purposes. Participants are
clustered using a matrix variate mixture model-based approach. Once clustered,
survival analysis is performed to demonstrate distinct lifetime outcomes for
individuals within each cluster. In summary, we process, cluster, and analyse a
subset of UK Biobank participants to show the protective effects from physical
activity on circadian disrupted individuals.
| stat |
A cross-center smoothness prior for variational Bayesian brain tissue
segmentation | Suppose one is faced with the challenge of tissue segmentation in MR images,
without annotators at their center to provide labeled training data. One option
is to go to another medical center for a trained classifier. Sadly, tissue
classifiers do not generalize well across centers due to voxel intensity shifts
caused by center-specific acquisition protocols. However, certain aspects of
segmentations, such as spatial smoothness, remain relatively consistent and can
be learned separately. Here we present a smoothness prior that is fit to
segmentations produced at another medical center. This informative prior is
presented to an unsupervised Bayesian model. The model clusters the voxel
intensities, such that it produces segmentations that are similarly smooth to
those of the other medical center. In addition, the unsupervised Bayesian model
is extended to a semi-supervised variant, which needs no visual interpretation
of clusters into tissues.
| stat |
Score-based Tests for Explaining Upper-Level Heterogeneity in Linear
Mixed Models | Cross-level interactions among fixed effects in linear mixed models (also
known as multilevel models) are often complicated by the variances stemming
from random effects and residuals. When these variances change across clusters,
tests of fixed effects (including cross-level interaction terms) are subject to
inflated Type I or Type II error. While the impact of variance
change/heterogeneity has been noticed in the literature, few methods have been
proposed to detect this heterogeneity in a simple, systematic way. In addition,
when heterogeneity among clusters is detected, researchers often wish to know
which clusters' variances differed from the others. In this study, we utilize a
recently-proposed family of score-based tests to distinguish between
cross-level interactions and heterogeneity in variance components, also
providing information about specific clusters that exhibit heterogeneity. These
score-based tests only require estimation of the null model (when variance
homogeneity is assumed to hold), and they have been previously applied to
psychometric models. We extend the tests to linear mixed models here, detailing
their implementation and performance when the data generating model is known.
We also include an empirical example illustrating the tests' use in practice.
| stat |
Global inducing point variational posteriors for Bayesian neural
networks and deep Gaussian processes | We derive the optimal approximate posterior over the top-layer weights in a
Bayesian neural network for regression, and show that it exhibits strong
dependencies on the lower-layer weights. We adapt this result to develop a
correlated approximate posterior over the weights at all layers in a Bayesian
neural network. We extend this approach to deep Gaussian processes, unifying
inference in the two model classes. Our approximate posterior uses learned
"global" inducing points, which are defined only at the input layer and
propagated through the network to obtain inducing inputs at subsequent layers.
By contrast, standard, "local", inducing point methods from the deep Gaussian
process literature optimize a separate set of inducing inputs at every layer,
and thus do not model correlations across layers. Our method gives
state-of-the-art performance for a variational Bayesian method, without data
augmentation or tempering, on CIFAR-10 of $86.7\%$.
| stat |
Doubly stochastic distributions of extreme events | The distribution of block maxima of sequences of independent and
identically-distributed random variables is used to model extreme values in
many disciplines. The traditional extreme value (EV) theory derives a
closed-form expression for the distribution of block maxima under asymptotic
assumptions, and is generally fitted using annual maxima or excesses over a
high threshold, thereby discarding a large fraction of the available
observations. The recently-introduced Metastatistical Extreme Value
Distribution (MEVD), a non-asymptotic formulation based on doubly stochastic
distributions, has been shown to offer several advantages compared to the
traditional EV theory. In particular, MEVD explicitly accounts for the
variability of the process generating the extreme values, and uses all the
available information to perform high-quantile inferences. Here we review the
derivation of the MEVD, analyzing its assumptions in detail, and show that its
general formulation includes other doubly stochastic approaches to extreme
value analysis that have been recently proposed.
| stat |
Inference in latent factor regression with clusterable features | Regression models, in which the observed features $X \in \R^p$ and the
response $Y \in \R$ depend, jointly, on a lower dimensional, unobserved, latent
vector $Z \in \R^K$, with $K< p$, are popular in a large array of applications,
and mainly used for predicting a response from correlated features. In
contrast, methodology and theory for inference on the regression coefficient
$\beta$ relating $Y$ to $Z$ are scarce, since typically the un-observable
factor $Z$ is hard to interpret. Furthermore, the determination of the
asymptotic variance of an estimator of $\beta$ is a long-standing problem, with
solutions known only in a few particular cases. To address some of these
outstanding questions, we develop inferential tools for $\beta$ in a class of
factor regression models in which the observed features are signed mixtures of
the latent factors. The model specifications are practically desirable, in a
large array of applications, render interpretability to the components of $Z$,
and are sufficient for parameter identifiability. Without assuming that the
number of latent factors $K$ or the structure of the mixture is known in
advance, we construct computationally efficient estimators of $\beta$, along
with estimators of other important model parameters. We benchmark the rate of
convergence of $\beta$ by first establishing its $\ell_2$-norm minimax lower
bound, and show that our proposed estimator is minimax-rate adaptive. Our main
contribution is the provision of a unified analysis of the component-wise
Gaussian asymptotic distribution of $\wh \beta$ and, especially, the derivation
of a closed form expression of its asymptotic variance, together with
consistent variance estimators. The resulting inferential tools can be used
when both $K$ and $p$ are independent of the sample size $n$, and when both, or
either, $p$ and $K$ vary with $n$, while allowing for $p > n$.
| stat |
On Multi-Armed Bandit Designs for Dose-Finding Clinical Trials | We study the problem of finding the optimal dosage in early stage clinical
trials through the multi-armed bandit lens. We advocate the use of the Thompson
Sampling principle, a flexible algorithm that can accommodate different types
of monotonicity assumptions on the toxicity and efficacy of the doses. For the
simplest version of Thompson Sampling, based on a uniform prior distribution
for each dose, we provide finite-time upper bounds on the number of sub-optimal
dose selections, which is unprecedented for dose-finding algorithms. Through a
large simulation study, we then show that variants of Thompson Sampling based
on more sophisticated prior distributions outperform state-of-the-art dose
identification algorithms in different types of dose-finding studies that occur
in phase I or phase I/II trials.
| stat |
Semi-Supervised Method using Gaussian Random Fields for Boilerplate
Removal in Web Browsers | Boilerplate removal refers to the problem of removing noisy content from a
webpage such as ads and extracting relevant content that can be used by various
services. This can be useful in several features in web browsers such as ad
blocking, accessibility tools such as read out loud, translation, summarization
etc. In order to create a training dataset to train a model for boilerplate
detection and removal, labeling or tagging webpage data manually can be tedious
and time consuming. Hence, a semi-supervised model, in which some of the
webpage elements are labeled manually and labels for others are inferred based
on some parameters, can be useful. In this paper we present a solution for
extraction of relevant content from a webpage that relies on semi-supervised
learning using Gaussian Random Fields. We first represent the webpage as a
graph, with text elements as nodes and the edge weights representing similarity
between nodes. After this, we label a few nodes in the graph using heuristics
and label the remaining nodes by a weighted measure of similarity to the
already labeled nodes. We describe the system architecture and a few
preliminary results on a dataset of webpages.
| stat |
Conditional Super Learner | In this article we consider the Conditional Super Learner (CSL), an algorithm
which selects the best model candidate from a library conditional on the
covariates. The CSL expands the idea of using cross-validation to select the
best model and merges it with meta learning. Here we propose a specific
algorithm that finds a local minimum to the problem posed, proof that it
converges at a rate faster than $O_p(n^{-1/4})$ and offers extensive empirical
evidence that it is an excellent candidate to substitute stacking or for the
analysis of Hierarchical problems.
| stat |
Probability Assessments of an Ice-Free Arctic: Comparing Statistical and
Climate Model Projections | The downward trend in the amount of Arctic sea ice has a wide range of
environmental and economic consequences including important effects on the pace
and intensity of global climate change. Based on several decades of satellite
data, we provide statistical forecasts of Arctic sea ice extent during the rest
of this century. The best fitting statistical model indicates that overall sea
ice coverage is declining at an increasing rate. By contrast, average
projections from the CMIP5 global climate models foresee a gradual slowing of
Arctic sea ice loss even in scenarios with high carbon emissions. Our
long-range statistical projections also deliver probability assessments of the
timing of an ice-free Arctic. These results indicate almost a 60 percent chance
of an effectively ice-free Arctic Ocean sometime during the 2030s -- much
earlier than the average projection from the global climate models.
| stat |
These Unprecedented Times: The Dynamic Pattern Of COVID-19 Deaths Around
The World | In this article, we deal with COVID-19 data to study the trend of the
epidemic at the global situation. Choosing the mortality rate as an appropriate
metric which measures the relative relation between the cumulative confirmed
cases and death cases, we utilize the modified kernel estimator for random
density function (Petersen and Muller, 2016) to obtain the density of the
mortality rate, and apply Frechet change point detection (Dubey and Muller,
2020) to check if there is any significant change on the dynamic pattern of
COVID-19 deaths. The analysis shows that the pattern of global COVID-19
mortality rate seems to have an obvious change at 104 days after first day when
the reported death cases exceed 30. Then we discuss the reasons of abrupt
change of mortality rate trend from aspects both subjective and objective.
| stat |
Resilient Monotone Sequential Maximization | Applications in machine learning, optimization, and control require the
sequential selection of a few system elements, such as sensors, data, or
actuators, to optimize the system performance across multiple time steps.
However, in failure-prone and adversarial environments, sensors get attacked,
data get deleted, and actuators fail. Thence, traditional sequential design
paradigms become insufficient and, in contrast, resilient sequential designs
that adapt against system-wide attacks, deletions, or failures become
important. In general, resilient sequential design problems are computationally
hard. Also, even though they often involve objective functions that are
monotone and (possibly) submodular, no scalable approximation algorithms are
known for their solution. In this paper, we provide the first scalable
algorithm, that achieves the following characteristics: system-wide resiliency,
i.e., the algorithm is valid for any number of denial-of-service attacks,
deletions, or failures; adaptiveness, i.e., at each time step, the algorithm
selects system elements based on the history of inflicted attacks, deletions,
or failures; and provable approximation performance, i.e., the algorithm
guarantees for monotone objective functions a solution close to the optimal. We
quantify the algorithm's approximation performance using a notion of curvature
for monotone (not necessarily submodular) set functions. Finally, we support
our theoretical analyses with simulated experiments, by considering a
control-aware sensor scheduling scenario, namely, sensing-constrained robot
navigation.
| stat |
Conditional Expectation Propagation | Expectation propagation (EP) is a powerful approximate inference algorithm.
However, a critical barrier in applying EP is that the moment matching in
message updates can be intractable. Handcrafting approximations is usually
tricky, and lacks generalizability. Importance sampling is very expensive.
While Laplace propagation provides a good solution, it has to run numerical
optimizations to find Laplace approximations in every update, which is still
quite inefficient. To overcome these practical barriers, we propose conditional
expectation propagation (CEP) that performs conditional moment matching given
the variables outside each message, and then takes expectation w.r.t the
approximate posterior of these variables. The conditional moments are often
analytical and much easier to derive. In the most general case, we can use
(fully) factorized messages to represent the conditional moments by quadrature
formulas. We then compute the expectation of the conditional moments via Taylor
approximations when necessary. In this way, our algorithm can always conduct
efficient, analytical fixed point iterations. Experiments on several popular
models for which standard EP is available or unavailable demonstrate the
advantages of CEP in both inference quality and computational efficiency.
| stat |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.