title
stringlengths 6
244
| abstract
stringlengths 19
5.09k
| label
stringclasses 10
values |
|---|---|---|
Near-optimal Individualized Treatment Recommendations | Individualized treatment recommendation (ITR) is an important analytic
framework for precision medicine. The goal is to assign proper treatments to
patients based on their individual characteristics. From the machine learning
perspective, the solution to an ITR problem can be formulated as a weighted
classification problem to maximize the average benefit that patients receive
from the recommended treatments. Several methods have been proposed for ITR in
both binary and multicategory treatment setups. In practice, one may prefer a
more flexible recommendation with multiple treatment options. This motivates us
to develop methods to obtain a set of near-optimal individualized treatment
recommendations alternative to each other, called alternative individualized
treatment recommendations (A-ITR). We propose two methods to estimate the
optimal A-ITR within the outcome weighted learning (OWL) framework. We show the
consistency of these methods and obtain an upper bound for the risk between the
theoretically optimal recommendation and the estimated one. We also conduct
simulation studies, and apply our methods to a real data set for Type 2
diabetic patients with injectable antidiabetic treatments. These numerical
studies have shown the usefulness of the proposed A-ITR framework. We develop a
R package aitr which can be found at https://github.com/menghaomiao/aitr.
| stat |
Improving Relational Regularized Autoencoders with Spherical Sliced
Fused Gromov Wasserstein | Relational regularized autoencoder (RAE) is a framework to learn the
distribution of data by minimizing a reconstruction loss together with a
relational regularization on the latent space. A recent attempt to reduce the
inner discrepancy between the prior and aggregated posterior distributions is
to incorporate sliced fused Gromov-Wasserstein (SFG) between these
distributions. That approach has a weakness since it treats every slicing
direction similarly, meanwhile several directions are not useful for the
discriminative task. To improve the discrepancy and consequently the relational
regularization, we propose a new relational discrepancy, named spherical sliced
fused Gromov Wasserstein (SSFG), that can find an important area of projections
characterized by a von Mises-Fisher distribution. Then, we introduce two
variants of SSFG to improve its performance. The first variant, named mixture
spherical sliced fused Gromov Wasserstein (MSSFG), replaces the vMF
distribution by a mixture of von Mises-Fisher distributions to capture multiple
important areas of directions that are far from each other. The second variant,
named power spherical sliced fused Gromov Wasserstein (PSSFG), replaces the vMF
distribution by a power spherical distribution to improve the sampling time in
high dimension settings. We then apply the new discrepancies to the RAE
framework to achieve its new variants. Finally, we conduct extensive
experiments to show that the new proposed autoencoders have favorable
performance in learning latent manifold structure, image generation, and
reconstruction.
| stat |
Distributed Computation for Marginal Likelihood based Model Choice | We propose a general method for distributed Bayesian model choice, using the
marginal likelihood, where each worker has access only to non-overlapping
subsets of the data. Our approach approximates the model evidence for the full
data set through Monte Carlo sampling from the posterior on every subset
generating a model evidence per subset. The model evidences per worker are then
consistently combined using a novel approach which corrects for the splitting
using summary statistics of the generated samples. This divide-and-conquer
approach allows Bayesian model choice in the large data setting, exploiting all
available information but limiting communication between workers. Our work
thereby complements the work on consensus Monte Carlo (Scott et al., 2016) by
explicitly enabling model choice. In addition, we show how the suggested
approach can be extended to model choice within a reversible jump setting that
explores multiple feature combinations within one run.
| stat |
Inference for High-dimensional Maximin Effects in Heterogeneous
Regression Models Using a Sampling Approach | Heterogeneity is an important feature of modern data sets and a central task
is to extract information from large-scale and heterogeneous data. In this
paper, we consider multiple high-dimensional linear models and adopt the
definition of maximin effect (Meinshausen, B{\"u}hlmann, AoS, 43(4),
1801--1830) to summarize the information contained in this heterogeneous model.
We define the maximin effect for a targeted population whose covariate
distribution is possibly different from that of the observed data. We further
introduce a ridge-type maximin effect to simultaneously account for reward
optimality and statistical stability. To identify the high-dimensional maximin
effect, we estimate the regression covariance matrix by a debiased estimator
and use it to construct the aggregation weights for the maximin effect. A main
challenge for statistical inference is that the estimated weights might have a
mixture distribution and the resulted maximin effect estimator is not
necessarily asymptotic normal. To address this, we devise a novel sampling
approach to construct the confidence interval for any linear contrast of
high-dimensional maximin effects. The coverage and precision properties of the
proposed confidence interval are studied. The proposed method is demonstrated
over simulations and a genetic data set on yeast colony growth under different
environments.
| stat |
Informative Bayesian Neural Network Priors for Weak Signals | Encoding domain knowledge into the prior over the high-dimensional weight
space of a neural network is challenging but essential in applications with
limited data and weak signals. Two types of domain knowledge are commonly
available in scientific applications: 1. feature sparsity (fraction of features
deemed relevant); 2. signal-to-noise ratio, quantified, for instance, as the
proportion of variance explained (PVE). We show how to encode both types of
domain knowledge into the widely used Gaussian scale mixture priors with
Automatic Relevance Determination. Specifically, we propose a new joint prior
over the local (i.e., feature-specific) scale parameters that encodes knowledge
about feature sparsity, and a Stein gradient optimization to tune the
hyperparameters in such a way that the distribution induced on the model's PVE
matches the prior distribution. We show empirically that the new prior improves
prediction accuracy, compared to existing neural network priors, on several
publicly available datasets and in a genetics application where signals are
weak and sparse, often outperforming even computationally intensive
cross-validation for hyperparameter tuning.
| stat |
Lurking Inferential Monsters? Quantifying bias in non-experimental
evaluations of school programs | This study examines whether unobserved factors substantially bias education
evaluations that rely on the Conditional Independence Assumption. We add 14 new
within-study comparisons to the literature, all from primary schools in
England. Across these 14 studies, we generate 42 estimates of selection bias
using a simple matching approach. A meta-analysis of the estimates suggests
that the distribution of underlying bias is centered around zero. The mean
absolute value of estimated bias is 0.03{\sigma}, and none of the 42 estimates
are larger than 0.11{\sigma}. Results are similar for math, reading and writing
outcomes. Overall, we find no evidence of substantial selection bias due to
unobserved characteristics. These findings may not generalise easily to other
settings or to more radical educational interventions, but they do suggest that
non-experimental approaches could play a greater role than they currently do in
generating reliable causal evidence for school education.
| stat |
Approximation of the Cell Under Test in Sliding Window Detection
Processes | Analysis of sliding window detection detection processes requires careful
consideration of the cell under test, which is an amplitude squared measurement
of the signal plus clutter in the complex domain. Some authors have suggested
that there is sufficient merit in the approximation of the cell under test by a
distributional model similar to that assumed for the clutter distribution.
Under such an assumption, the development of Neyman-Pearson detectors is
facilitated. The purpose of the current paper is to demonstrate, in a modern
maritime surveillance radar context, that such an approximation is invalid.
| stat |
A statistical method for estimating the no-observed-adverse-event-level | In toxicological risk assessment the benchmark dose (BMD) is recommended
instead of the no-observed-adverse effect-level (NOAEL). Still a simple test
procedure to estimate NOAEL is proposed here, explaining its advantages and
disadvantages. Versatile applicability is illustrated using four different data
examples of selected in vivo toxicity bioassays.
| stat |
The Calculus of M-estimation in R with geex | M-estimation, or estimating equation, methods are widely applicable for point
estimation and asymptotic inference. In this paper, we present an R package
that can find roots and compute the empirical sandwich variance estimator for
any set of user-specified, unbiased estimating equations. Examples from the
M-estimation primer by Stefanski and Boos (2002) demonstrate use of the
software. The package also includes a framework for finite sample variance
corrections and a website with an extensive collection of tutorials.
| stat |
Optimal Confidence Regions for the Multinomial Parameter | Construction of tight confidence regions and intervals is central to
statistical inference and decision making. This paper develops new theory
showing minimum average volume confidence regions for categorical data. More
precisely, consider an empirical distribution $\widehat{\boldsymbol{p}}$
generated from $n$ iid realizations of a random variable that takes one of $k$
possible values according to an unknown distribution $\boldsymbol{p}$. This is
analogous to a single draw from a multinomial distribution. A confidence region
is a subset of the probability simplex that depends on
$\widehat{\boldsymbol{p}}$ and contains the unknown $\boldsymbol{p}$ with a
specified confidence. This paper shows how one can construct minimum average
volume confidence regions, answering a long standing question. We also show the
optimality of the regions directly translates to optimal confidence intervals
of linear functionals such as the mean, implying sample complexity and regret
improvements for adaptive machine learning algorithms.
| stat |
Distributionally Robust Counterfactual Risk Minimization | This manuscript introduces the idea of using Distributionally Robust
Optimization (DRO) for the Counterfactual Risk Minimization (CRM) problem.
Tapping into a rich existing literature, we show that DRO is a principled tool
for counterfactual decision making. We also show that well-established
solutions to the CRM problem like sample variance penalization schemes are
special instances of a more general DRO problem. In this unifying framework, a
variety of distributionally robust counterfactual risk estimators can be
constructed using various probability distances and divergences as uncertainty
measures. We propose the use of Kullback-Leibler divergence as an alternative
way to model uncertainty in CRM and derive a new robust counterfactual
objective. In our experiments, we show that this approach outperforms the
state-of-the-art on four benchmark datasets, validating the relevance of using
other uncertainty measures in practical applications.
| stat |
Implementing multiple imputation for missing data in longitudinal
studies when models are not feasible: A tutorial on the random hot deck
approach | Objective: Researchers often use model-based multiple imputation to handle
missing at random data to minimize bias while making the best use of all
available data. However, there are sometimes constraints within the data that
make model-based imputation difficult and may result in implausible values. In
these contexts, we describe how to use random hot deck imputation to allow for
plausible multiple imputation in longitudinal studies.
Study Design and Setting: We illustrate random hot deck multiple imputation
using The Childhood Health, Activity, and Motor Performance School Study
Denmark (CHAMPS-DK), a prospective cohort study that measured weekly sports
participation for 1700 Danish schoolchildren. We matched records with missing
data to several observed records, generated probabilities for matched records
using observed data, and sampled from these records based on the probability of
each occurring. Because imputed values are generated randomly, multiple
complete datasets can be created and analyzed similar to model-based multiple
imputation.
Conclusion: Multiple imputation using random hot deck imputation is an
alternative method when model-based approaches are infeasible, specifically
where there are constraints within and between covariates.
| stat |
Learning partially ranked data based on graph regularization | Ranked data appear in many different applications, including voting and
consumer surveys. There often exhibits a situation in which data are partially
ranked. Partially ranked data is thought of as missing data. This paper
addresses parameter estimation for partially ranked data under a (possibly)
non-ignorable missing mechanism. We propose estimators for both complete
rankings and missing mechanisms together with a simple estimation procedure.
Our estimation procedure leverages a graph regularization in conjunction with
the Expectation-Maximization algorithm. Our estimation procedure is
theoretically guaranteed to have the convergence properties. We reduce a
modeling bias by allowing a non-ignorable missing mechanism. In addition, we
avoid the inherent complexity within a non-ignorable missing mechanism by
introducing a graph regularization. The experimental results demonstrate that
the proposed estimators work well under non-ignorable missing mechanisms.
| stat |
Global estimation and scenario-based projections of sex ratio at birth
and missing female births using a Bayesian hierarchical time series mixture
model | The sex ratio at birth (SRB) is defined as the ratio of male to female live
births. The SRB imbalance in parts of the world over the past several decades
is a direct consequence of sex-selective abortion, driven by the co-existence
of son preference, readily available technology of prenatal sex determination,
and fertility decline. Estimation and projection of the degree of SRB imbalance
is complicated because of variability in SRB reference levels and because of
the uncertainty associated with SRB observations. We develop Bayesian
hierarchical time series mixture models for SRB estimation and scenario-based
projections for all countries from 1950 to 2100. We model the SRB regional and
national reference levels, and the fluctuation around national reference
levels. We identify countries at risk of SRB imbalances and model both (i) the
absence or presence of sex ratio transitions in such countries and, if present,
(ii) the transition process. The transition model of SRB imbalance captures
three stages (increase, stagnation and convergence back to SRB baselines). The
model identifies countries with statistical evidence of SRB inflation in a
fully Bayesian approach. The scenario-based SRB projections are based on the
sex ratio transition model with varying assumptions regarding the occurrence of
a sex ratio transition in at-risk countries. Projections are used to quantify
the future burden of missing female births due to sex-selective abortions under
different scenarios.
| stat |
Financial Application of Extended Residual Coherence | Residual coherence is a graphical tool for selecting potential second-order
interaction terms as functions of a single time series and its lags. This paper
extends the notion of residual coherence to account for interaction terms of
multiple time series. Moreover, an alternative criterion, integrated spectrum,
is proposed to facilitate this graphical selection.
A financial market application shows that new insights can be gained
regarding implied market volatility.
| stat |
Estimating animal utilization distributions from multiple data types: a
joint spatio-temporal point process framework | Models of the spatial distribution of animals provide useful tools to help
ecologists quantify species-environment relationships, and they are
increasingly being used to help determine the impacts of climate and habitat
changes on species. While high-quality survey-style data with known effort are
sometimes available, often researchers have multiple datasets of varying
quality and type. In particular, collections of sightings made by citizen
scientists are becoming increasingly common, with no information typically
provided on their observer effort. Many standard modelling approaches ignore
observer effort completely, which can severely bias estimates of an animal's
distribution. Combining sightings data from observers who followed different
protocols is challenging. Any differences in observer skill, spatial effort,
and the detectability of the animals across space all need to be accounted for.
To achieve this, we build upon the recent advancements made in integrative
species distribution models and present a novel marked spatio-temporal point
process framework for estimating the utilization distribution (UD) of the
individuals of a highly mobile species. We show that in certain settings, we
can also use the framework to combine the UDs from the sampled individuals to
estimate the species' distribution. We combine the empirical results from a
simulation study with the implications outlined in a causal directed acyclic
graph to identify the necessary assumptions required for our framework to
control for observer effort when it is unknown. We then apply our framework to
combine multiple datasets collected on the endangered Southern Resident Killer
Whales, to estimate their monthly effort-corrected space-use.
| stat |
Predictive Complexity Priors | Specifying a Bayesian prior is notoriously difficult for complex models such
as neural networks. Reasoning about parameters is made challenging by the
high-dimensionality and over-parameterization of the space. Priors that seem
benign and uninformative can have unintuitive and detrimental effects on a
model's predictions. For this reason, we propose predictive complexity priors:
a functional prior that is defined by comparing the model's predictions to
those of a reference model. Although originally defined on the model outputs,
we transfer the prior to the model parameters via a change of variables. The
traditional Bayesian workflow can then proceed as usual. We apply our
predictive complexity prior to high-dimensional regression, reasoning over
neural network depth, and sharing of statistical strength for few-shot
learning.
| stat |
Personalized acute stress classification from physiological signals with
neural processes | Objective: A person's affective state has known relationships to
physiological processes which can be measured by wearable sensors. However,
while there are general trends those relationships can be person-specific. This
work proposes using neural processes as a way to address individual
differences.
Methods: Stress classifiers built from classic machine learning models and
from neural processes are compared on two datasets using
leave-one-participant-out cross-validation. The neural processes models are
contextualized on data from a brief period of a particular person's recording.
Results: The neural processes models outperformed the standard machine
learning models, and had the best performance when using periods of stress and
baseline as context. Contextual points chosen from other participants led to
lower performance.
Conclusion: Neural processes can learn to adapt to person-specific
physiological sensor data. There are a wide range of affective and medical
applications for which this model could prove useful.
| stat |
Latent Association Mining in Binary Data | We consider the problem of identifying stable sets of mutually associated
features in moderate or high-dimensional binary data. In this context we
develop and investigate a method called Latent Association Mining for Binary
Data (LAMB). The LAMB method is based on a simple threshold model in which the
observed binary values represent a random thresholding of a latent continuous
vector that may have a complex association structure. We consider a measure of
latent association that quantifies association in the latent continuous vector
without bias due to the random thresholding. The LAMB method uses an iterative
testing based search procedure to identify stable sets of mutually associated
features. We compare the LAMB method with several competing methods on
artificial binary-valued datasets and two real count-valued datasets. The LAMB
method detects meaningful associations in these datasets. In the case of the
count-valued datasets, associations detected by the LAMB method are based only
on information about whether the counts are zero or non-zero, and is
competitive with methods that have access to the full count data.
| stat |
Beyond prediction: A framework for inference with variational
approximations in mixture models | Variational inference is a popular method for estimating model parameters and
conditional distributions in hierarchical and mixed models, which arise
frequently in many settings in the health, social, and biological sciences.
Variational inference in a frequentist context works by approximating
intractable conditional distributions with a tractable family and optimizing
the resulting lower bound on the log-likelihood. The variational objective
function is typically less computationally intensive to optimize than the true
likelihood, enabling scientists to fit rich models even with extremely large
datasets. Despite widespread use, little is known about the general theoretical
properties of estimators arising from variational approximations to the
log-likelihood, which hinders their use in inferential statistics. In this
paper we connect such estimators to profile M-estimation, which enables us to
provide regularity conditions for consistency and asymptotic normality of
variational estimators. Our theory also motivates three methodological
improvements to variational inference: estimation of the asymptotic
model-robust covariance matrix, a one-step correction that improves estimator
efficiency, and an empirical assessment of consistency. We evaluate the
proposed results using simulation studies and data on marijuana use from the
National Longitudinal Study of Youth.
| stat |
MCMC for a hyperbolic Bayesian inverse problem in traffic flow modelling | As work on hyperbolic Bayesian inverse problems remains rare in the
literature, we explore empirically the sampling challenges these offer which
have to do with shock formation in the solution of the PDE. Furthermore, we
provide a unified statistical model to estimate using motorway data both
boundary conditions and fundamental diagram parameters in LWR, a well known
motorway traffic flow model. This allows us to provide a traffic flow density
estimation method that is shown to be superior to two methods found in the
traffic flow literature. Finally, we highlight how \emph{Population Parallel
Tempering} - a modification of Parallel Tempering - is a scalable method that
can increase the mixing speed of the sampler by a factor of 10.
| stat |
On the use of approximate Bayesian computation Markov chain Monte Carlo
with inflated tolerance and post-correction | Approximate Bayesian computation allows for inference of complicated
probabilistic models with intractable likelihoods using model simulations. The
Markov chain Monte Carlo implementation of approximate Bayesian computation is
often sensitive to the tolerance parameter: low tolerance leads to poor mixing
and large tolerance entails excess bias. We consider an approach using a
relatively large tolerance for the Markov chain Monte Carlo sampler to ensure
its sufficient mixing, and post-processing the output leading to estimators for
a range of finer tolerances. We introduce an approximate confidence interval
for the related post-corrected estimators, and propose an adaptive approximate
Bayesian computation Markov chain Monte Carlo, which finds a `balanced'
tolerance level automatically, based on acceptance rate optimisation. Our
experiments show that post-processing based estimators can perform better than
direct Markov chain targetting a fine tolerance, that our confidence intervals
are reliable, and that our adaptive algorithm leads to reliable inference with
little user specification.
| stat |
Confidence intervals for parameters in high-dimensional sparse vector
autoregression | Vector autoregression (VAR) models are widely used to analyze the
interrelationship between multiple variables over time. Estimation and
inference for the transition matrices of VAR models are crucial for
practitioners to make decisions in fields such as economics and finance.
However, when the number of variables is larger than the sample size, it
remains a challenge to perform statistical inference of the model parameters.
In this article, we propose the de-biased Lasso and two bootstrap de-biased
Lasso methods to construct confidence intervals for the elements of the
transition matrices of high-dimensional VAR models. We show that the proposed
methods are asymptotically valid under appropriate sparsity and other
regularity conditions. To implement our methods, we develop feasible and
parallelizable algorithms, which save a large amount of computation required by
the nodewise Lasso and bootstrap. A simulation study illustrates that our
methods perform well in finite samples. Finally, we apply our methods to
analyze the price data of stocks in the S&P 500 index in 2019. We find that
some stocks, such as the largest producer of gold in the world, Newmont
Corporation, have significant predictive power over the most stocks.
| stat |
Model Monitoring and Dynamic Model Selection in Travel Time-series
Forecasting | Accurate travel products price forecasting is a highly desired feature that
allows customers to take informed decisions about purchases, and companies to
build and offer attractive tour packages. Thanks to machine learning (ML), it
is now relatively cheap to develop highly accurate statistical models for price
time-series forecasting. However, once models are deployed in production, it is
their monitoring, maintenance and improvement which carry most of the costs and
difficulties over time. We introduce a data-driven framework to continuously
monitor and maintain deployed time-series forecasting models' performance, to
guarantee stable performance of travel products price forecasting models. Under
a supervised learning approach, we predict the errors of time-series
forecasting models over time, and use this predicted performance measure to
achieve both model monitoring and maintenance. We validate the proposed method
on a dataset of 18K time-series from flight and hotel prices collected over two
years and on two public benchmarks.
| stat |
An Early Warning Approach to Monitor COVID-19 Activity with Multiple
Digital Traces in Near Real-Time | Non-pharmaceutical interventions (NPIs) have been crucial in curbing COVID-19
in the United States (US). Consequently, relaxing NPIs through a phased
re-opening of the US amid still-high levels of COVID-19 susceptibility could
lead to new epidemic waves. This calls for a COVID-19 early warning system.
Here we evaluate multiple digital data streams as early warning indicators of
increasing or decreasing state-level US COVID-19 activity between January and
June 2020. We estimate the timing of sharp changes in each data stream using a
simple Bayesian model that calculates in near real-time the probability of
exponential growth or decay. Analysis of COVID-19-related activity on social
network microblogs, Internet searches, point-of-care medical software, and a
metapopulation mechanistic model, as well as fever anomalies captured by smart
thermometer networks, shows exponential growth roughly 2-3 weeks prior to
comparable growth in confirmed COVID-19 cases and 3-4 weeks prior to comparable
growth in COVID-19 deaths across the US over the last 6 months. We further
observe exponential decay in confirmed cases and deaths 5-6 weeks after
implementation of NPIs, as measured by anonymized and aggregated human mobility
data from mobile phones. Finally, we propose a combined indicator for
exponential growth in multiple data streams that may aid in developing an early
warning system for future COVID-19 outbreaks. These efforts represent an
initial exploratory framework, and both continued study of the predictive power
of digital indicators as well as further development of the statistical
approach are needed.
| stat |
Gaussian processes with multidimensional distribution inputs via optimal
transport and Hilbertian embedding | In this work, we investigate Gaussian Processes indexed by multidimensional
distributions. While directly constructing radial positive definite kernels
based on the Wasserstein distance has been proven to be possible in the
unidimensional case, such constructions do not extend well to the
multidimensional case as we illustrate here. To tackle the problem of defining
positive definite kernels between multivariate distributions based on optimal
transport, we appeal instead to Hilbert space embeddings relying on optimal
transport maps to a reference distribution, that we suggest to take as a
Wasserstein barycenter. We characterize in turn radial positive definite
kernels on Hilbert spaces, and show that the covariance parameters of virtually
all parametric families of covariance functions are microergodic in the case of
(infinite-dimensional) Hilbert spaces. We also investigate statistical
properties of our suggested positive definite kernels on multidimensional
distributions, with a focus on consistency when a population Wasserstein
barycenter is replaced by an empirical barycenter and additional explicit
results in the special case of Gaussian distributions. Finally, we study the
Gaussian process methodology based on our suggested positive definite kernels
in regression problems with multidimensional distribution inputs, on simulation
data stemming both from synthetic examples and from a mechanical engineering
test case.
| stat |
Local linear tie-breaker designs | Tie-breaker experimental designs are hybrids of Randomized Control Trials
(RCTs) and Regression Discontinuity Designs (RDDs) in which subjects with
moderate scores are placed in an RCT while subjects with extreme scores are
deterministically assigned to the treatment or control group. The design
maintains the benefits of randomization for causal estimation while avoiding
the possibility of excluding the most deserving recipients from the treatment
group. The causal effect estimator for a tie-breaker design can be estimated by
fitting local linear regressions for both the treatment and control group, as
is typically done for RDDs. We study the statistical efficiency of such local
linear regression-based causal estimators as a function of $\Delta$, the radius
of the interval in which treatment randomization occurs. In particular, we
determine the efficiency of the estimator as a function of $\Delta$ for a
fixed, arbitrary bandwidth under the assumption of a uniform assignment
variable. To generalize beyond uniform assignment variables and asymptotic
regimes, we also demonstrate on the Angrist and Lavy (1999) classroom size
dataset that prior to conducting an experiment, an experimental designer can
estimate the efficiency for various experimental radii choices by using Monte
Carlo as long as they have access to the distribution of the assignment
variable. For both uniform and triangular kernels, we show that increasing the
radius of randomized experiment interval will increase the efficiency until the
radius is the size of the local-linear regression bandwidth, after which no
additional efficiency benefits are conferred.
| stat |
Toroidal AutoEncoder | Enforcing distributions of latent variables in neural networks is an active
subject. It is vital in all kinds of generative models, where we want to be
able to interpolate between points in the latent space, or sample from it.
Modern generative AutoEncoders (AE) like WAE, SWAE, CWAE add a regularizer to
the standard (deterministic) AE, which allows to enforce Gaussian distribution
in the latent space. Enforcing different distributions, especially
topologically nontrivial, might bring some new interesting possibilities, but
this subject seems unexplored so far.
This article proposes a new approach to enforce uniform distribution on
d-dimensional torus. We introduce a circular spring loss, which enforces
minibatch points to be equally spaced and satisfy cyclic boundary conditions.
As example of application we propose multiple-path morphing. Minimal distance
geodesic between two points in uniform distribution on latent space of angles
becomes a line, however, torus topology allows us to choose such lines in
alternative ways, going through different edges of $[-\pi,\pi]^d$.
Further applications to explore can be for example trying to learn real-life
topologically nontrivial spaces of features, like rotations to automatically
recognize 2D rotation of an object in picture by training on relative angles,
or even 3D rotations by additionally using spherical features - this way
morphing should be close to object rotation.
| stat |
Geometry-Aware Maximum Likelihood Estimation of Intrinsic Dimension | The existing approaches to intrinsic dimension estimation usually are not
reliable when the data are nonlinearly embedded in the high dimensional space.
In this work, we show that the explicit accounting to geometric properties of
unknown support leads to the polynomial correction to the standard maximum
likelihood estimate of intrinsic dimension for flat manifolds. The proposed
algorithm (GeoMLE) realizes the correction by regression of standard MLEs based
on distances to nearest neighbors for different sizes of neighborhoods.
Moreover, the proposed approach also efficiently handles the case of nonuniform
sampling of the manifold. We perform numerous experiments on different
synthetic and real-world datasets. The results show that our algorithm achieves
state-of-the-art performance, while also being computationally efficient and
robust to noise in the data.
| stat |
High Dimensional Classification for Spatially Dependent Data with
Application to Neuroimaging | Discriminating patients with Alzheimer's disease (AD) from healthy subjects
is a crucial task in the research of Alzheimer's disease. The task can be
potentially achieved by linear discriminant analysis (LDA), which is one of the
most classical and popular classification techniques. However, the
classification problem becomes challenging for LDA because of the
high-dimensionally and the spatial dependency of the brain imaging data. To
address the challenges, researchers have proposed various ways to generalize
LDA into high-dimensional context in recent years. However, these existing
methods did not reach any consensus on how to incorporate spatially dependent
structure. In light of the current needs and limitations, we propose a new
classification method, named as Penalized Maximum Likelihood Estimation LDA
(PMLE-LDA). The proposed method uses $Mat\acute{e}rn$ covariance function to
describe the spatial correlation of brain regions. Additionally, PMLE is
designed to model the sparsity of high-dimensional features. The spatial
location information is used to address the singularity of the covariance.
Tapering technique is introduced to reduce computational burden. We show in
theory that the proposed method can not only provide consistent results of
parameter estimation and feature selection, but also generate an asymptotically
optimal classifier driven by high dimensional data with specific spatially
dependent structure. Finally, the method is validated through simulations and
an application into ADNI data for classifying Alzheimer's patients.
| stat |
Stochastic Epidemic Models inference and diagnosis with Poisson Random
Measure Data Augmentation | We present a new Bayesian inference method for compartmental models that
takes into account the intrinsic stochasticity of the process. We show how to
formulate a SIR-type Markov jump process as the solution of a stochastic
differential equation with respect to a Poisson Random Measure (PRM), and how
to simulate the process trajectory deterministically from a parameter value and
a PRM realisation.
This forms the basis of our Data Augmented MCMC, which consists in augmenting
parameter space with the unobserved PRM value. The resulting simple
Metropolis-Hastings sampler acts as an efficient simulation-based inference
method, that can easily be transferred from model to model. Compared with a
recent Data Augmentation method based on Gibbs sampling of individual infection
histories, PRM-augmented MCMC scales much better with epidemic size and is far
more flexible.
PRM-augmented MCMC also yields a posteriori estimates of the PRM, that
represent process stochasticity, and which can be used to validate the model.
If the model is good, the posterior distribution should exhibit no pattern and
be close to the PRM prior distribution. We illustrate this by fitting a
non-seasonal model to some simulated seasonal case count data. Applied to the
Zika epidemic of 2013 in French Polynesia, our approach shows that a simple
SEIR model cannot correctly reproduce both the initial sharp increase in the
number of cases as well as the final proportion of seropositive.
PRM-augmentation thus provides a coherent story for Stochastic Epidemic Model
inference, where explicitly inferring process stochasticity helps with model
validation.
| stat |
A comparison of some conformal quantile regression methods | We compare two recently proposed methods that combine ideas from conformal
inference and quantile regression to produce locally adaptive and marginally
valid prediction intervals under sample exchangeability (Romano et al., 2019;
Kivaranovic et al., 2019). First, we prove that these two approaches are
asymptotically efficient in large samples, under some additional assumptions.
Then we compare them empirically on simulated and real data. Our results
demonstrate that the method in Romano et al. (2019) typically yields tighter
prediction intervals in finite samples. Finally, we discuss how to tune these
procedures by fixing the relative proportions of observations used for training
and conformalization.
| stat |
Particle Filter Bridge Interpolation | Auto encoding models have been extensively studied in recent years. They
provide an efficient framework for sample generation, as well as for analysing
feature learning. Furthermore, they are efficient in performing interpolations
between data-points in semantically meaningful ways. In this paper, we build
further on a previously introduced method for generating canonical, dimension
independent, stochastic interpolations. Here, the distribution of interpolation
paths is represented as the distribution of a bridge process constructed from
an artificial random data generating process in the latent space, having the
prior distribution as its invariant distribution. As a result the stochastic
interpolation paths tend to reside in regions of the latent space where the
prior has high mass. This is a desirable feature since, generally, such areas
produce semantically meaningful samples. In this paper, we extend the bridge
process method by introducing a discriminator network that accurately
identifies areas of high latent representation density. The discriminator
network is incorporated as a change of measure of the underlying bridge process
and sampling of interpolation paths is implemented using sequential Monte
Carlo. The resulting sampling procedure allows for greater variability in
interpolation paths and stronger drift towards areas of high data density.
| stat |
Covariate-Powered Empirical Bayes Estimation | We study methods for simultaneous analysis of many noisy experiments in the
presence of rich covariate information. The goal of the analyst is to optimally
estimate the true effect underlying each experiment. Both the noisy
experimental results and the auxiliary covariates are useful for this purpose,
but neither data source on its own captures all the information available to
the analyst. In this paper, we propose a flexible plug-in empirical Bayes
estimator that synthesizes both sources of information and may leverage any
black-box predictive model. We show that our approach is within a constant
factor of minimax for a simple data-generating model. Furthermore, we establish
robust convergence guarantees for our method that hold under considerable
generality, and exhibit promising empirical performance on both real and
simulated data.
| stat |
Bayesian sample size determination using commensurate priors to leverage
pre-experimental data | This paper develops Bayesian sample size formulae for experiments comparing
two groups. We assume the experimental data will be analysed in the Bayesian
framework, where pre-experimental information from multiple sources can be
represented into robust priors. In particular, such robust priors account for
preliminary belief about the pairwise commensurability between parameters that
underpin the historical and new experiments, to permit flexible borrowing of
information. Averaged over the probability space of the new experimental data,
appropriate sample sizes are found according to criteria that control certain
aspects of the posterior distribution, such as the coverage probability or
length of a defined density region. Our Bayesian methodology can be applied to
circumstances where the common variance in the new experiment is known or
unknown. Exact solutions are available based on most of the criteria considered
for Bayesian sample size determination, while a search procedure is described
in cases for which there are no closed-form expressions. We illustrate the
application of our Bayesian sample size formulae in the setting of designing a
clinical trial. Hypothetical data examples, motivated by a rare-disease trial
with elicitation of expert prior opinion, and a comprehensive performance
evaluation of the proposed methodology are presented.
| stat |
Anomaly Detection in Large Scale Networks with Latent Space Models | We develop a real-time anomaly detection algorithm for directed activity on
large, sparse networks. We model the propensity for future activity using a
dynamic logistic model with interaction terms for sender- and receiver-specific
latent factors in addition to sender- and receiver-specific popularity scores;
deviations from this underlying model constitute potential anomalies. Latent
nodal attributes are estimated via a variational Bayesian approach and may
change over time, representing natural shifts in network activity. Estimation
is augmented with a case-control approximation to take advantage of the
sparsity of the network and reduces computational complexity from $O(N^2)$ to
$O(E)$, where $N$ is the number of nodes and $E$ is the number of observed
edges. We run our algorithm on network event records collected from an
enterprise network of over 25,000 computers and are able to identify a red team
attack with half the detection rate required of the model without latent
interaction terms.
| stat |
TrajectoryNet: A Dynamic Optimal Transport Network for Modeling Cellular
Dynamics | It is increasingly common to encounter data from dynamic processes captured
by static cross-sectional measurements over time, particularly in biomedical
settings. Recent attempts to model individual trajectories from this data use
optimal transport to create pairwise matchings between time points. However,
these methods cannot model continuous dynamics and non-linear paths that
entities can take in these systems. To address this issue, we establish a link
between continuous normalizing flows and dynamic optimal transport, that allows
us to model the expected paths of points over time. Continuous normalizing
flows are generally under constrained, as they are allowed to take an arbitrary
path from the source to the target distribution. We present TrajectoryNet,
which controls the continuous paths taken between distributions to produce
dynamic optimal transport. We show how this is particularly applicable for
studying cellular dynamics in data from single-cell RNA sequencing (scRNA-seq)
technologies, and that TrajectoryNet improves upon recently proposed static
optimal transport-based models that can be used for interpolating cellular
distributions.
| stat |
Fast structure learning with modular regularization | Estimating graphical model structure from high-dimensional and undersampled
data is a fundamental problem in many scientific fields. Existing approaches,
such as GLASSO, latent variable GLASSO, and latent tree models, suffer from
high computational complexity and may impose unrealistic sparsity priors in
some cases. We introduce a novel method that leverages a newly discovered
connection between information-theoretic measures and structured latent factor
models to derive an optimization objective which encourages modular structures
where each observed variable has a single latent parent. The proposed method
has linear stepwise computational complexity w.r.t. the number of observed
variables. Our experiments on synthetic data demonstrate that our approach is
the only method that recovers modular structure better as the dimensionality
increases. We also use our approach for estimating covariance structure for a
number of real-world datasets and show that it consistently outperforms
state-of-the-art estimators at a fraction of the computational cost. Finally,
we apply the proposed method to high-resolution fMRI data (with more than 10^5
voxels) and show that it is capable of extracting meaningful patterns.
| stat |
A Bayesian Approach for De-duplication in the Presence of Relational
Data | In this paper we study the impact of combining profile and network data in a
de-duplication setting. We also assess the influence of a range of prior
distributions on the linkage structure, including our proposal. Our proposed
prior makes it straightforward to specify prior believes and naturally enforces
the microclustering property. Furthermore, we explore stochastic gradient
Hamiltonian Monte Carlo methods as a faster alternative to obtain samples for
the network parameters. Our methodology is evaluated using the RLdata500 data,
which is a popular dataset in the record linkage literature.
| stat |
Unlocking GOES: A Statistical Framework for Quantifying the Evolution of
Convective Structure in Tropical Cyclones | Tropical cyclones (TCs) rank among the most costly natural disasters in the
United States, and accurate forecasts of track and intensity are critical for
emergency response. Intensity guidance has improved steadily but slowly, as
processes which drive intensity change are not fully understood. Because most
TCs develop far from land-based observing networks, geostationary satellite
imagery is critical to monitor these storms. However, these complex data can be
challenging to analyze in real time, and off-the-shelf machine learning
algorithms have limited applicability on this front due to their ``black box''
structure. This study presents analytic tools that quantify convective
structure patterns in infrared satellite imagery for over-ocean TCs, yielding
lower-dimensional but rich representations that support analysis and
visualization of how these patterns evolve during rapid intensity change. The
proposed ORB feature suite targets the global Organization, Radial structure,
and Bulk morphology of TCs. By combining ORB and empirical orthogonal
functions, we arrive at an interpretable and rich representation of convective
structure patterns that serve as inputs to machine learning methods. This study
uses the logistic lasso, a penalized generalized linear model, to relate
predictors to rapid intensity change. Using ORB alone, binary classifiers
identifying the presence (versus absence) of such intensity change events can
achieve accuracy comparable to classifiers using environmental predictors
alone, with a combined predictor set improving classification accuracy in some
settings. More complex nonlinear machine learning methods did not perform
better than the linear logistic lasso model for current data.
| stat |
A flexible Bayesian framework to estimate age- and cause-specific child
mortality over time from sample registration data | In order to implement disease-specific interventions in young age groups,
policy makers in low- and middle-income countries require timely and accurate
estimates of age- and cause-specific child mortality. High quality data is not
available in settings where these interventions are most needed, but there is a
push to create sample registration systems that collect detailed mortality
information. Current methods that estimate mortality from this data employ
multistage frameworks without rigorous statistical justification that
separately estimate all-cause and cause-specific mortality and are not
sufficiently adaptable to capture important features of the data. We propose a
flexible Bayesian modeling framework to estimate age- and cause-specific child
mortality from sample registration data. We provide a theoretical justification
for the framework, explore its properties via simulation, and use it to
estimate mortality trends using data from the Maternal and Child Health
Surveillance System in China.
| stat |
A Hidden Markov Model Based Unsupervised Algorithm for Sleep/Wake
Identification Using Actigraphy | Actigraphy is widely used in sleep studies but lacks a universal unsupervised
algorithm for sleep/wake identification. In this study, we proposed a Hidden
Markov Model (HMM) based unsupervised algorithm that can automatically and
effectively infer sleep/wake states. It is an individualized data-driven
approach that analyzes actigraphy from each individual respectively to learn
activity characteristics and further separate sleep and wake states. We used
Actiwatch and polysomnography (PSG) data from 43 individuals in the
Multi-Ethnic Study of Atherosclerosis to evaluate the performance of our
method. Epoch-by-epoch comparisons were made between our HMM algorithm and that
embedded in the Actiwatch software (AS). The percent agreement between HMM and
PSG was 85.7%, and that between AS and PSG was 84.7%. Positive predictive
values for sleep epochs were 85.6% and 84.6% for HMM and AS, respectively, and
95.5% and 85.6% for wake epochs. Both methods have similar performance and tend
to overestimate sleep and underestimate wake compared to PSG. Our HMM approach
is able to quantify the variability in activity counts that allow us to
differentiate relatively active and sedentary individuals: individuals with
higher estimated variabilities tend to show more frequent sedentary behaviors.
In conclusion, our unsupervised data-driven HMM algorithm achieves slightly
better performance compared to the commonly used algorithm in the Actiwatch
software. HMM can help expand the application of actigraphy in large-scale
studies and in cases where intrusive PSG is hard to acquire or unavailable. In
addition, the estimated HMM parameters can characterize individual activity
patterns that can be utilized for further analysis.
| stat |
Nonparametric Empirical Bayes Estimation on Heterogeneous Data | The simultaneous estimation of many parameters $\eta_i$, based on a
corresponding set of observations $x_i$, for $i=1,\ldots, n$, is a key research
problem that has received renewed attention in the high-dimensional setting.
%The classic example involves estimating a vector of normal means $\mu_i$
subject to a fixed variance term $\sigma^2$. However, Many practical situations
involve heterogeneous data $(x_i, \theta_i)$ where $\theta_i$ is a known
nuisance parameter. Effectively pooling information across samples while
correctly accounting for heterogeneity presents a significant challenge in
large-scale estimation problems. We address this issue by introducing the
"Nonparametric Empirical Bayes Smoothing Tweedie" (NEST) estimator, which
efficiently estimates $\eta_i$ and properly adjusts for heterogeneity %by
approximating the marginal density of the data $f_{\theta_i}(x_i)$ and applying
this density to via a generalized version of Tweedie's formula. NEST is capable
of handling a wider range of settings than previously proposed heterogeneous
approaches as it does not make any parametric assumptions on the prior
distribution of $\eta_i$. The estimation framework is simple but general enough
to accommodate any member of the exponential family of distributions. %; a
thorough study of the normal means problem subject to heterogeneous variances
is presented to illustrate the proposed framework. Our theoretical results show
that NEST is asymptotically optimal, while simulation studies show that it
outperforms competing methods, with substantial efficiency gains in many
settings. The method is demonstrated on a data set measuring the performance
gap in math scores between socioeconomically advantaged and disadvantaged
students in K-12 schools.
| stat |
Bayesian Sparse Factor Analysis with Kernelized Observations | Multi-view problems can be faced with latent variable models since they are
able to find low-dimensional projections that fairly capture the correlations
among the multiple views that characterise each datum. On the other hand,
high-dimensionality and non-linear issues are traditionally handled by kernel
methods, inducing a (non)-linear function between the latent projection and the
data itself. However, they usually come with scalability issues and exposition
to overfitting. Here, we propose merging both approaches into single model so
that we can exploit the best features of multi-view latent models and kernel
methods and, moreover, overcome their limitations.
In particular, we combine probabilistic factor analysis with what we refer to
as kernelized observations, in which the model focuses on reconstructing not
the data itself, but its relationship with other data points measured by a
kernel function. This model can combine several types of views (kernelized or
not), and it can handle heterogeneous data and work in semi-supervised
settings. Additionally, by including adequate priors, it can provide compact
solutions for the kernelized observations -- based in a automatic selection of
Bayesian Relevance Vectors (RVs) -- and can include feature selection
capabilities. Using several public databases, we demonstrate the potential of
our approach (and its extensions) w.r.t. common multi-view learning models such
as kernel canonical correlation analysis or manifold relevance determination.
| stat |
Adversarial Variational Bayes Methods for Tweedie Compound Poisson Mixed
Models | The Tweedie Compound Poisson-Gamma model is routinely used for modeling
non-negative continuous data with a discrete probability mass at zero. Mixed
models with random effects account for the covariance structure related to the
grouping hierarchy in the data. An important application of Tweedie mixed
models is pricing the insurance policies, e.g. car insurance. However, the
intractable likelihood function, the unknown variance function, and the
hierarchical structure of mixed effects have presented considerable challenges
for drawing inferences on Tweedie. In this study, we tackle the Bayesian
Tweedie mixed-effects models via variational inference approaches. In
particular, we empower the posterior approximation by implicit models trained
in an adversarial setting. To reduce the variance of gradients, we
reparameterize random effects, and integrate out one local latent variable of
Tweedie. We also employ a flexible hyper prior to ensure the richness of the
approximation. Our method is evaluated on both simulated and real-world data.
Results show that the proposed method has smaller estimation bias on the random
effects compared to traditional inference methods including MCMC; it also
achieves a state-of-the-art predictive performance, meanwhile offering a richer
estimation of the variance function.
| stat |
The Effectiveness of Multitask Learning for Phenotyping with Electronic
Health Records Data | Electronic phenotyping is the task of ascertaining whether an individual has
a medical condition of interest by analyzing their medical record and is
foundational in clinical informatics. Increasingly, electronic phenotyping is
performed via supervised learning. We investigate the effectiveness of
multitask learning for phenotyping using electronic health records (EHR) data.
Multitask learning aims to improve model performance on a target task by
jointly learning additional auxiliary tasks and has been used in disparate
areas of machine learning. However, its utility when applied to EHR data has
not been established, and prior work suggests that its benefits are
inconsistent. We present experiments that elucidate when multitask learning
with neural nets improves performance for phenotyping using EHR data relative
to neural nets trained for a single phenotype and to well-tuned logistic
regression baselines. We find that multitask neural nets consistently
outperform single-task neural nets for rare phenotypes but underperform for
relatively more common phenotypes. The effect size increases as more auxiliary
tasks are added. Moreover, multitask learning reduces the sensitivity of neural
nets to hyperparameter settings for rare phenotypes. Last, we quantify
phenotype complexity and find that neural nets trained with or without
multitask learning do not improve on simple baselines unless the phenotypes are
sufficiently complex.
| stat |
A hierarchical spatio-temporal model to analyze relative risk variations
of COVID-19: a focus on Spain, Italy and Germany | The novel coronavirus disease (COVID-19) has spread rapidly across the world
in a short period of time and with a heterogeneous pattern. Understanding the
underlying temporal and spatial dynamics in the spread of COVID-19 can result
in informed and timely public health policies. In this paper, we use a
spatio-temporal stochastic model to explain the temporal and spatial variations
in the daily number of new confirmed cases in Spain, Italy and Germany from
late February to mid September 2020. Using a hierarchical Bayesian framework,
we found that the temporal trend of the epidemic in the three countries rapidly
reached their peaks and slowly started to decline at the beginning of April and
then increased and reached their second maximum in August. However decline and
increase of the temporal trend seems to be sharper in Spain and smoother in
Germany. The spatial heterogeneity of the relative risk of COVID-19 in Spain is
also more pronounced than Italy and Germany.
| stat |
An end-to-end Differentially Private Latent Dirichlet Allocation Using a
Spectral Algorithm | We provide an end-to-end differentially private spectral algorithm for
learning LDA, based on matrix/tensor decompositions, and establish theoretical
guarantees on utility/consistency of the estimated model parameters. The
spectral algorithm consists of multiple algorithmic steps, named as "{edges}",
to which noise could be injected to obtain differential privacy. We identify
\emph{subsets of edges}, named as "{configurations}", such that adding noise to
all edges in such a subset guarantees differential privacy of the end-to-end
spectral algorithm. We characterize the sensitivity of the edges with respect
to the input and thus estimate the amount of noise to be added to each edge for
any required privacy level. We then characterize the utility loss for each
configuration as a function of injected noise. Overall, by combining the
sensitivity and utility characterization, we obtain an end-to-end
differentially private spectral algorithm for LDA and identify the
corresponding configuration that outperforms others in any specific regime. We
are the first to achieve utility guarantees under the required level of
differential privacy for learning in LDA. Overall our method systematically
outperforms differentially private variational inference.
| stat |
Discrete factor analysis | In this paper, we present a method for factor analysis of discrete data. This
is accomplished by fitting a dependent Poisson model with a factor structure.
To be able to analyze ordinal data, we also consider a truncated Poisson
distribution. We try to find the model with the lowest AIC by employing a
forward selection procedure. The probability to find the correct model is
investigated in a simulation study. Moreover, we heuristically derive the
corresponding asymptotic probabilities. An empirical study is also included.
| stat |
Quantifying Perceptual Distortion of Adversarial Examples | Recent work has shown that additive threat models, which only permit the
addition of bounded noise to the pixels of an image, are insufficient for fully
capturing the space of imperceivable adversarial examples. For example, small
rotations and spatial transformations can fool classifiers, remain
imperceivable to humans, but have large additive distance from the original
images. In this work, we leverage quantitative perceptual metrics like LPIPS
and SSIM to define a novel threat model for adversarial attacks.
To demonstrate the value of quantifying the perceptual distortion of
adversarial examples, we present and employ a unifying framework fusing
different attack styles. We first prove that our framework results in images
that are unattainable by attack styles in isolation. We then perform
adversarial training using attacks generated by our framework to demonstrate
that networks are only robust to classes of adversarial perturbations they have
been trained against, and combination attacks are stronger than any of their
individual components. Finally, we experimentally demonstrate that our combined
attacks retain the same perceptual distortion but induce far higher
misclassification rates when compared against individual attacks.
| stat |
Classification of chemical compounds based on the correlation between
\textit{in vitro} gene expression profiles | Toxicity evaluation of chemical compounds has traditionally relied on animal
experiments;however, the demand for non-animal-based prediction methods for
toxicology of compounds is increasing worldwide. Our aim was to provide a
classification method for compounds based on \textit{in vitro} gene expression
profiles. The \textit{in vitro} gene expression data analyzed in the present
study was obtained from our previous study. The data concerned nine compounds
typically employed in chemical management.We used agglomerative hierarchical
clustering to classify the compounds;however, there was a statistical
difficulty to be overcome.We needed to properly extract RNAs for clustering
from more than 30,000 RNAs. In order to overcome this difficulty, we introduced
a combinatorial optimization problem with respect to both gene expression
levels and the correlation between gene expression profiles. Then, the
simulated annealing algorithm was used to obtain a good solution for the
problem. As a result, the nine compounds were divided into two groups using
1,000 extracted RNAs. Our proposed methodology enables read-across, one of the
frameworks for predicting toxicology, based on \textit{in vitro} gene
expression profiles.
| stat |
Gain with no Pain: Efficient Kernel-PCA by Nystr\"om Sampling | In this paper, we propose and study a Nystr\"om based approach to efficient
large scale kernel principal component analysis (PCA). The latter is a natural
nonlinear extension of classical PCA based on considering a nonlinear feature
map or the corresponding kernel. Like other kernel approaches, kernel PCA
enjoys good mathematical and statistical properties but, numerically, it scales
poorly with the sample size. Our analysis shows that Nystr\"om sampling greatly
improves computational efficiency without incurring any loss of statistical
accuracy. While similar effects have been observed in supervised learning, this
is the first such result for PCA. Our theoretical findings, which are also
illustrated by numerical results, are based on a combination of analytic and
concentration of measure techniques. Our study is more broadly motivated by the
question of understanding the interplay between statistical and computational
requirements for learning.
| stat |
Alternative Analysis Methods for Time to Event Endpoints under
Non-proportional Hazards: A Comparative Analysis | The log-rank test is most powerful under proportional hazards (PH). In
practice, non-PH patterns are often observed in clinical trials, such as in
immuno-oncology; therefore, alternative methods are needed to restore the
efficiency of statistical testing. Three categories of testing methods were
evaluated, including weighted log-rank tests, Kaplan-Meier curve-based tests
(including weighted Kaplan-Meier and Restricted Mean Survival Time, RMST), and
combination tests (including Breslow test, Lee's combo test, and MaxCombo
test). Nine scenarios representing the PH and various non-PH patterns were
simulated. The power, type I error, and effect estimates of each method were
compared. In general, all tests control type I error well. There is not a
single most powerful test across all scenarios. In the absence of prior
knowledge regarding the PH or non-PH patterns, the MaxCombo test is relatively
robust across patterns. Since the treatment effect changes overtime under
non-PH, the overall profile of the treatment effect may not be represented
comprehensively based on a single measure. Thus, multiple measures of the
treatment effect should be pre-specified as sensitivity analyses to evaluate
the totality of the data.
| stat |
Sparse Travel Time Estimation from Streaming Data | We address two shortcomings in online travel time estimation methods for
congested urban traffic. The first shortcoming is related to the determination
of the number of mixture modes, which can change dynamically, within day and
from day to day. The second shortcoming is the wide-spread use of Gaussian
probability densities as mixture components. Gaussian densities fail to capture
the positive skew in travel time distributions and, consequently, large numbers
of mixture components are needed for reasonable fitting accuracy when applied
as mixture components. They also assign positive probabilities to negative
travel times. To address these issues, this paper derives a mixture
distribution with Gamma component densities, which are asymmetric and supported
on the positive numbers. We use sparse estimation techniques to ensure
parsimonious models and propose a generalization of Gamma mixture densities
using Mittag-Leffler functions, which provides enhanced fitting flexibility and
improved parsimony. In order to accommodate within-day variability and allow
for online implementation of the proposed methodology (i.e., fast computations
on streaming travel time data), we introduce a recursive algorithm which
efficiently updates the fitted distribution whenever new data become available.
Experimental results using real-world travel time data illustrate the efficacy
of the proposed methods.
| stat |
Bayesian Regression Using a Prior on the Model Fit: The R2-D2 Shrinkage
Prior | Prior distributions for high-dimensional linear regression require specifying
a joint distribution for the unobserved regression coefficients, which is
inherently difficult. We instead propose a new class of shrinkage priors for
linear regression via specifying a prior first on the model fit, in particular,
the coefficient of determination, and then distributing through to the
coefficients in a novel way. The proposed method compares favourably to
previous approaches in terms of both concentration around the origin and tail
behavior, which leads to improved performance both in posterior contraction and
in empirical performance. The limiting behavior of the proposed prior is $1/x$,
both around the origin and in the tails. This behavior is optimal in the sense
that it simultaneously lies on the boundary of being an improper prior both in
the tails and around the origin. None of the existing shrinkage priors obtain
this behavior in both regions simultaneously. We also demonstrate that our
proposed prior leads to the same near-minimax posterior contraction rate as the
spike-and-slab prior.
| stat |
LR-GLM: High-Dimensional Bayesian Inference Using Low-Rank Data
Approximations | Due to the ease of modern data collection, applied statisticians often have
access to a large set of covariates that they wish to relate to some observed
outcome. Generalized linear models (GLMs) offer a particularly interpretable
framework for such an analysis. In these high-dimensional problems, the number
of covariates is often large relative to the number of observations, so we face
non-trivial inferential uncertainty; a Bayesian approach allows coherent
quantification of this uncertainty. Unfortunately, existing methods for
Bayesian inference in GLMs require running times roughly cubic in parameter
dimension, and so are limited to settings with at most tens of thousand
parameters. We propose to reduce time and memory costs with a low-rank
approximation of the data in an approach we call LR-GLM. When used with the
Laplace approximation or Markov chain Monte Carlo, LR-GLM provides a full
Bayesian posterior approximation and admits running times reduced by a full
factor of the parameter dimension. We rigorously establish the quality of our
approximation and show how the choice of rank allows a tunable
computational-statistical trade-off. Experiments support our theory and
demonstrate the efficacy of LR-GLM on real large-scale datasets.
| stat |
Two-Stage Penalized Regression Screening to Detect Biomarker-Treatment
Interactions in Randomized Clinical Trials | High-dimensional biomarkers such as genomics are increasingly being measured
in randomized clinical trials. Consequently, there is a growing interest in
developing methods that improve the power to detect biomarker-treatment
interactions. We adapt recently proposed two-stage interaction detecting
procedures in the setting of randomized clinical trials. We also propose a new
stage 1 multivariate screening strategy using ridge regression to account for
correlations among biomarkers. For this multivariate screening, we prove the
asymptotic between-stage independence, required for family-wise error rate
control, under biomarker-treatment independence. Simulation results show that
in various scenarios, the ridge regression screening procedure can provide
substantially greater power than the traditional one-biomarker-at-a-time
screening procedure in highly correlated data. We also exemplify our approach
in two real clinical trial data applications.
| stat |
The Restricted Isometry of ReLU Networks: Generalization through Norm
Concentration | While regression tasks aim at interpolating a relation on the entire input
space, they often have to be solved with a limited amount of training data.
Still, if the hypothesis functions can be sketched well with the data, one can
hope for identifying a generalizing model.
In this work, we introduce with the Neural Restricted Isometry Property
(NeuRIP) a uniform concentration event, in which all shallow $\mathrm{ReLU}$
networks are sketched with the same quality. To derive the sample complexity
for achieving NeuRIP, we bound the covering numbers of the networks in the
Sub-Gaussian metric and apply chaining techniques. In case of the NeuRIP event,
we then provide bounds on the expected risk, which hold for networks in any
sublevel set of the empirical risk. We conclude that all networks with
sufficiently small empirical risk generalize uniformly.
| stat |
pivmet: Pivotal Methods for Bayesian Relabelling and k-Means Clustering | The identification of groups' prototypes, i.e. elements of a dataset that
represent different groups of data points, may be relevant to the tasks of
clustering, classification and mixture modeling. The R package pivmet presented
in this paper includes different methods for extracting pivotal units from a
dataset. One of the main applications of pivotal methods is a Markov Chain
Monte Carlo (MCMC) relabelling procedure to solve the label switching in
Bayesian estimation of mixture models. Each method returns posterior estimates,
and a set of graphical tools for visualizing the output. The package offers
JAGS and Stan sampling procedures for Gaussian mixtures, and allows for
user-defined priors' parameters. The package also provides functions to perform
consensus clustering based on pivotal units, which may allow to improve
classical techniques (e.g. k-means) by means of a careful seeding. The paper
provides examples of applications to both real and simulated datasets.
| stat |
Analytically Embedding Differential Equation Constraints into Least
Squares Support Vector Machines using the Theory of Functional Connections | Differential equations (DEs) are used as numerical models to describe
physical phenomena throughout the field of engineering and science, including
heat and fluid flow, structural bending, and systems dynamics. While there are
many other techniques for finding approximate solutions to these equations,
this paper looks to compare the application of the Theory of Functional
Connections (TFC) with one based on least-squares support vector machines
(LS-SVM). The TFC method uses a constrained expression, an expression that
always satisfies the DE constraints, which transforms the process of solving a
DE into solving an unconstrained optimization problem that is ultimately solved
via least-squares (LS). In addition to individual analysis, the two methods are
merged into a new methodology, called constrained SVMs (CSVM), by incorporating
the LS-SVM method into the TFC framework to solve unconstrained problems.
Numerical tests are conducted on four sample problems: One first order linear
ordinary differential equation (ODE), one first order nonlinear ODE, one second
order linear ODE, and one two-dimensional linear partial differential equation
(PDE). Using the LS-SVM method as a benchmark, a speed comparison is made for
all the problems by timing the training period, and an accuracy comparison is
made using the maximum error and mean squared error on the training and test
sets. In general, TFC is shown to be slightly faster (by an order of magnitude
or less) and more accurate (by multiple orders of magnitude) than the LS-SVM
and CSVM approaches.
| stat |
Informed reversible jump algorithms | Incorporating information about the target distribution in proposal
mechanisms generally increases the efficiency of Markov chain Monte Carlo
algorithms, when compared with those based on naive random walks. Hamiltonian
Monte Carlo represents a successful example of fixed-dimensional algorithms
incorporating gradient information. In trans-dimensional algorithms, Green
(2003) recommended to generate the parameter proposals during model switches
from normal distributions with informative means and covariance matrices. These
proposal distributions can be viewed as asymptotic approximations to the
parameter distributions, where the limit is with regard to the sample size.
Models are typically proposed using uninformed uniform distributions. In this
paper, we build on the approach of Zanella (2020) for discrete spaces to
incorporate information about neighbouring models. We rely on approximations to
posterior model probabilities that are asymptotically exact. We prove that, in
some scenarios, the samplers combining this approach with that of Green (2003)
behave like those that use the exact model probabilities and sample from the
parameter distributions, in the large sample regime. We show that the
implementation of the proposed samplers is straightforward in some cases. The
methodology is applied to a real-data example. The code is available online.
| stat |
Discussion of Kallus (2020) and Mo, Qi, and Liu (2020): New Objectives
for Policy Learning | We discuss the thought-provoking new objective functions for policy learning
that were proposed in "More efficient policy learning via optimal retargeting"
by Nathan Kallus and "Learning optimal distributionally robust individualized
treatment rules" by Weibin Mo, Zhengling Qi, and Yufeng Liu. We show that it is
important to take the curvature of the value function into account when working
within the retargeting framework, and we introduce two ways to do so. We also
describe more efficient approaches for leveraging calibration data when
learning distributionally robust policies.
| stat |
A Bayesian Feature Allocation Model for Identification of Cell
Subpopulations Using Cytometry Data | A Bayesian feature allocation model (FAM) is presented for identifying cell
subpopulations based on multiple samples of cell surface or intracellular
marker expression level data obtained by cytometry by time of flight (CyTOF).
Cell subpopulations are characterized by differences in expression patterns of
makers, and individual cells are clustered into the subpopulations based on the
patterns of their observed expression levels. A finite Indian buffet process is
used to model subpopulations as latent features, and a model-based method based
on these latent feature subpopulations is used to construct cell clusters
within each sample. Non-ignorable missing data due to technical artifacts in
mass cytometry instruments are accounted for by defining a static missing data
mechanism. In contrast to conventional cell clustering methods based on
observed marker expression levels that are applied separately to different
samples, the FAM based method can be applied simultaneously to multiple
samples, and can identify important cell subpopulations likely to be missed by
conventional clustering. The proposed FAM based method is applied to jointly
analyze three datasets, generated by CyTOF, to study natural killer (NK) cells.
Because the subpopulations identified by the FAM may define novel NK cell
subsets, this statistical analysis may provide useful information about the
biology of NK cells and their potential role in cancer immunotherapy which may
lead, in turn, to development of improved cellular therapies. Simulation
studies of the proposed method's behavior under two cases of known
subpopulations also are presented, followed by analysis of the CyTOF NK cell
surface marker data.
| stat |
Modeling pages left blank in university examination: A resolution in
higher education process | Trees are the main sources of paper production, in most of the cases, as far
as the intellectual usages are concerned. However, our planet is lacking in
that particular natural resource due to rapid growth of population,
urbanization, and increased pollution, more importantly non-judicial
utilization of such kind. Indian education sectors (schools, colleges,
universities) utilize a major part in consumption of papers as a classical
practice for conducting examinations and other documentation activities. Our
attempt in this article is to investigate and provide an optimal estimate of
the number of pages actually required in answer booklet in higher education
sector. Truncated Poisson distribution is found to be the best fit for the data
on number of pages left blank in an answer booklet after conduction of semester
end examinations. To predict the outcome based on various factors such as,
lines per pages, words per line, types of examinations etc. suitable regression
modeling is performed. A real data set, collected over a period of one month,
is been analyzed to illustrate the methods and conclusion is accomplished in
the direction of cost reduction, saving of papers, and in turn, logical uses of
natural resource to protect environmental interests.
| stat |
Structured Graph Learning Via Laplacian Spectral Constraints | Learning a graph with a specific structure is essential for interpretability
and identification of the relationships among data. It is well known that
structured graph learning from observed samples is an NP-hard combinatorial
problem. In this paper, we first show that for a set of important graph
families it is possible to convert the structural constraints of structure into
eigenvalue constraints of the graph Laplacian matrix. Then we introduce a
unified graph learning framework, lying at the integration of the spectral
properties of the Laplacian matrix with Gaussian graphical modeling that is
capable of learning structures of a large class of graph families. The proposed
algorithms are provably convergent and practically amenable for large-scale
semi-supervised and unsupervised graph-based learning tasks. Extensive
numerical experiments with both synthetic and real data sets demonstrate the
effectiveness of the proposed methods. An R package containing code for all the
experimental results is available at
https://cran.r-project.org/package=spectralGraphTopology.
| stat |
Bayesian Model Averaging for Causality Estimation and its Approximation
based on Gaussian Scale Mixture Distributions | In the estimation of the causal effect under linear Structural Causal Models
(SCMs), it is common practice to first identify the causal structure, estimate
the probability distributions, and then calculate the causal effect. However,
if the goal is to estimate the causal effect, it is not necessary to fix a
single causal structure or probability distributions. In this paper, we first
show from a Bayesian perspective that it is Bayes optimal to weight (average)
the causal effects estimated under each model rather than estimating the causal
effect under a fixed single model. This idea is also known as Bayesian model
averaging. Although the Bayesian model averaging is optimal, as the number of
candidate models increases, the weighting calculations become computationally
hard. We develop an approximation to the Bayes optimal estimator by using
Gaussian scale mixture distributions.
| stat |
Bayesian Algorithm Execution: Estimating Computable Properties of
Black-box Functions Using Mutual Information | In many real world problems, we want to infer some property of an expensive
black-box function f, given a budget of T function evaluations. One example is
budget constrained global optimization of f, for which Bayesian optimization is
a popular method. Other properties of interest include local optima, level
sets, integrals, or graph-structured information induced by f. Often, we can
find an algorithm A to compute the desired property, but it may require far
more than T queries to execute. Given such an A, and a prior distribution over
f, we refer to the problem of inferring the output of A using T evaluations as
Bayesian Algorithm Execution (BAX). To tackle this problem, we present a
procedure, InfoBAX, that sequentially chooses queries that maximize mutual
information with respect to the algorithm's output. Applying this to Dijkstra's
algorithm, for instance, we infer shortest paths in synthetic and real-world
graphs with black-box edge costs. Using evolution strategies, we yield variants
of Bayesian optimization that target local, rather than global, optima. On
these problems, InfoBAX uses up to 500 times fewer queries to f than required
by the original algorithm. Our method is closely connected to other Bayesian
optimal experimental design procedures such as entropy search methods and
optimal sensor placement using Gaussian processes.
| stat |
Adaptive Inference for Change Points in High-Dimensional Data | In this article, we propose a class of test statistics for a change point in
the mean of high-dimensional independent data. Our test integrates the
U-statistic based approach in a recent work by \cite{hdcp} and the $L_q$-norm
based high-dimensional test in \cite{he2018}, and inherits several appealing
features such as being tuning parameter free and asymptotic independence for
test statistics corresponding to even $q$s. A simple combination of test
statistics corresponding to several different $q$s leads to a test with
adaptive power property, that is, it can be powerful against both sparse and
dense alternatives. On the estimation front, we obtain the convergence rate of
the maximizer of our test statistic standardized by sample size when there is
one change-point in mean and $q=2$, and propose to combine our tests with a
wild binary segmentation (WBS) algorithm to estimate the change-point number
and locations when there are multiple change-points. Numerical comparisons
using both simulated and real data demonstrate the advantage of our adaptive
test and its corresponding estimation method.
| stat |
Development and validation of computable Phenotype to Identify and
Characterize Kidney Health in Adult Hospitalized Patients | Background: Acute kidney injury (AKI) is a common complication in
hospitalized patients and a common cause for chronic kidney disease (CKD) and
increased hospital cost and mortality. By timely detection of AKI and AKI
progression, effective preventive or therapeutic measures could be offered.
This study aims to develop and validate an electronic phenotype to identify
patients with CKD and AKI. Methods: A database with electronic health records
data from a retrospective study cohort of 84,352 hospitalized adults was
created. This repository includes demographics, comorbidities, vital signs,
laboratory values, medications, diagnoses and procedure codes for all index
admission, 12 months prior and 12 months follow-up encounters. We developed
algorithms to identify CKD and AKI based on the Kidney Disease: Improving
Global Outcomes (KDIGO) criteria. To measure diagnostic performance of the
algorithms, clinician experts performed clinical adjudication of AKI and CKD on
300 selected cases. Results: Among 149,136 encounters, identified CKD by
medical history was 12% which increased to 16% using creatinine criteria. Among
130,081 encounters with sufficient data for AKI phenotyping 21% had AKI. The
comparison of CKD phenotyping algorithm to manual chart review yielded PPV of
0.87, NPV of 0.99, sensitivity of 0.99, and specificity of 0.89. The comparison
of AKI phenotyping algorithm to manual chart review yielded PPV of 0.99, NPV of
0.95 , sensitivity 0.98, and specificity 0.98. Conclusions: We developed
phenotyping algorithms that yielded very good performance in identification of
patients with CKD and AKI in validation cohort. This tool may be useful in
identifying patients with kidney disease in a large population, in assessing
the quality and value of care in such patients.
| stat |
Provable Noisy Sparse Subspace Clustering using Greedy Neighbor
Selection: A Coherence-Based Perspective | Sparse subspace clustering (SSC) using greedy-based neighbor selection, such
as matching pursuit (MP) and orthogonal matching pursuit (OMP), has been known
as a popular computationally-efficient alternative to the conventional
L1-minimization based methods. Under deterministic bounded noise corruption, in
this paper we derive coherence-based sufficient conditions guaranteeing correct
neighbor identification using MP/OMP. Our analyses exploit the maximum/minimum
inner product between two noisy data points subject to a known upper bound on
the noise level. The obtained sufficient condition clearly reveals the impact
of noise on greedy-based neighbor recovery. Specifically, it asserts that, as
long as noise is sufficiently small so that the resultant perturbed residual
vectors stay close to the desired subspace, both MP and OMP succeed in
returning a correct neighbor subset. A striking finding is that, when the
ground truth subspaces are well-separated from each other and noise is not
large, MP-based iterations, while enjoying lower algorithmic complexity, yield
smaller perturbation of residuals, thereby better able to identify correct
neighbors and, in turn, achieving higher global data clustering accuracy.
Extensive numerical experiments are used to corroborate our theoretical study.
| stat |
Learning Schatten--von Neumann Operators | We study the learnability of a class of compact operators known as
Schatten--von Neumann operators. These operators between infinite-dimensional
function spaces play a central role in a variety of applications in learning
theory and inverse problems. We address the question of sample complexity of
learning Schatten-von Neumann operators and provide an upper bound on the
number of measurements required for the empirical risk minimizer to generalize
with arbitrary precision and probability, as a function of class parameter $p$.
Our results give generalization guarantees for regression of
infinite-dimensional signals from infinite-dimensional data. Next, we adapt the
representer theorem of Abernethy \emph{et al.} to show that empirical risk
minimization over an a priori infinite-dimensional, non-compact set, can be
converted to a convex finite dimensional optimization problem over a compact
set. In summary, the class of $p$-Schatten--von Neumann operators is probably
approximately correct (PAC)-learnable via a practical convex program for any $p
< \infty$.
| stat |
Order-Independent Structure Learning of Multivariate Regression Chain
Graphs | This paper deals with multivariate regression chain graphs (MVR CGs), which
were introduced by Cox and Wermuth [3,4] to represent linear causal models with
correlated errors. We consider the PC-like algorithm for structure learning of
MVR CGs, which is a constraint-based method proposed by Sonntag and Pe\~{n}a in
[18]. We show that the PC-like algorithm is order-dependent, in the sense that
the output can depend on the order in which the variables are given. This
order-dependence is a minor issue in low-dimensional settings. However, it can
be very pronounced in high-dimensional settings, where it can lead to highly
variable results. We propose two modifications of the PC-like algorithm that
remove part or all of this order-dependence. Simulations under a variety of
settings demonstrate the competitive performance of our algorithms in
comparison with the original PC-like algorithm in low-dimensional settings and
improved performance in high-dimensional settings.
| stat |
A Case Study of Space-time Performance Comparison of Wind Turbines on a
Wind Farm | This paper presents an academia-industry joint case study, which was
conducted to quantify and compare multi-year changes in power production
performance of multiple turbines scattered over a mid-size wind farm. This
analysis is referred to as a space-time performance comparison. One key aspect
in power performance analysis is to have the wind and environmental inputs
controlled for. This research employs, in a sequential fashion, two principal
modeling components to exercise tight control of multiple input conditions -- a
covariate matching method, followed by a Gaussian process model-based
functional comparison. The analysis method is applied to a wind farm that
houses 66 turbines on a moderately complex terrain. The power production and
environmental data span nearly four years, during which period the turbines
have gone through multiple technical upgrades. The space-time analysis presents
a quantitative and global picture showing how turbines differ relative to each
other as well as how each of them changes over time.
| stat |
Amortised Learning by Wake-Sleep | Models that employ latent variables to capture structure in observed data lie
at the heart of many current unsupervised learning algorithms, but exact
maximum-likelihood learning for powerful and flexible latent-variable models is
almost always intractable. Thus, state-of-the-art approaches either abandon the
maximum-likelihood framework entirely, or else rely on a variety of variational
approximations to the posterior distribution over the latents. Here, we propose
an alternative approach that we call amortised learning. Rather than computing
an approximation to the posterior over latents, we use a wake-sleep Monte-Carlo
strategy to learn a function that directly estimates the maximum-likelihood
parameter updates. Amortised learning is possible whenever samples of latents
and observations can be simulated from the generative model, treating the model
as a "black box". We demonstrate its effectiveness on a wide range of complex
models, including those with latents that are discrete or supported on
non-Euclidean spaces.
| stat |
Active learning for efficiently training emulators of computationally
expensive mathematical models | An emulator is a fast-to-evaluate statistical approximation of a detailed
mathematical model (simulator). When used in lieu of simulators, emulators can
expedite tasks that require many repeated evaluations, such as sensitivity
analyses, policy optimization, model calibration, and value-of-information
analyses. Emulators are developed using the output of simulators at specific
input values (design points). Developing an emulator that closely approximates
the simulator can require many design points, which becomes computationally
expensive. We describe a self-terminating active learning algorithm to
efficiently develop emulators tailored to a specific emulation task, and
compare it with algorithms that optimize geometric criteria (random latin
hypercube sampling and maximum projection designs) and other active learning
algorithms (treed Gaussian Processes that optimize typical active learning
criteria). We compared the algorithms' root mean square error (RMSE) and
maximum absolute deviation from the simulator (MAX) for seven benchmark
functions and in a prostate cancer screening model. In the empirical analyses,
in simulators with greatly-varying smoothness over the input domain, active
learning algorithms resulted in emulators with smaller RMSE and MAX for the
same number of design points. In all other cases, all algorithms performed
comparably. The proposed algorithm attained satisfactory performance in all
analyses, had smaller variability than the treed Gaussian Processes (it is
deterministic), and, on average, had similar or better performance as the treed
Gaussian Processes in 6 out of 7 benchmark functions and in the prostate cancer
model.
| stat |
A Bayesian Decision Tree Algorithm | Bayesian Decision Trees are known for their probabilistic interpretability.
However, their construction can sometimes be costly. In this article we present
a general Bayesian Decision Tree algorithm applicable to both regression and
classification problems. The algorithm does not apply Markov Chain Monte Carlo
and does not require a pruning step. While it is possible to construct a
weighted probability tree space we find that one particular tree, the
greedy-modal tree (GMT), explains most of the information contained in the
numerical examples. This approach seems to perform similarly to Random Forests.
| stat |
Calibration of Spatial Forecasts from Citizen Science Urban Air
Pollution Data with Sparse Recurrent Neural Networks | With their continued increase in coverage and quality, data collected from
personal air quality monitors has become an increasingly valuable tool to
complement existing public health monitoring system over urban areas. However,
the potential of using such `citizen science data' for automatic early warning
systems is hampered by the lack of models able to capture the high-resolution,
nonlinear spatio-temporal features stemming from local emission sources such as
traffic, residential heating and commercial activities. In this work, we
propose a machine learning approach to forecast high-frequency spatial fields
which has two distinctive advantages from standard neural network methods in
time: 1) sparsity of the neural network via a spike-and-slab prior, and 2) a
small parametric space. The introduction of stochastic neural networks
generates additional uncertainty, and in this work we propose a fast approach
for forecast calibration, both marginal and spatial. We focus on assessing
exposure to urban air pollution in San Francisco, and our results suggest an
improvement of 35.7% in the mean squared error over standard time series
approach with a calibrated forecast for up to 5 days.
| stat |
Gait Change Detection Using Parameters Generated from Microsoft Kinect
Coordinates | This paper describes a method to convert Microsoft Kinect coordinates into
gait parameters in order to detect a person's gait change. The proposed method
can help quantify the progress of physical therapy. Microsoft Kinect, a popular
platform for video games, was used to generate 25 joints to form a human
skeleton, and then the proposed method converted the coordinates of selected
Kinect joints into gait parameters such as spine tilt, hip tilt, and shoulder
tilt, which were tracked over time. Sample entropy measure was then applied to
quantify the variability of each gait parameter. Male and female subjects
walked a three-meter path multiple times in initial experiments, and their
walking patterns were recorded via the proposed Kinect device through the
frontal plane. Time series of the gait parameters were generated for subjects
with and without knee braces. Sample entropy was used to transform these time
series into numerical values for comparison of these two conditions.
| stat |
The Elliptical Ornstein-Uhlenbeck Process | We introduce the elliptical Ornstein-Uhlenbeck (OU) process, which is a
generalisation of the well-known univariate OU process to bivariate time
series. This process maps out elliptical stochastic oscillations over time in
the complex plane, which are observed in many applications of coupled bivariate
time series. The appeal of the model is that elliptical oscillations are
generated using one simple first order SDE, whereas alternative models require
more complicated vectorised or higher order SDE representations. The second
useful feature is that parameter estimation can be performed robustly in the
frequency domain using only the modelled and observed power spectral density,
without having to model and compute cross spectra of individual time series
components. We determine properties of the model including the conditions for
stationarity, and the geometrical structure of the elliptical oscillations. We
demonstrate the utility of the model by measuring periodic and elliptical
properties of Earth's polar motion.
| stat |
Nonparametric Quantile Regression for Homogeneity Pursuit in Panel Data
Models | Many panel data have the latent subgroup effect on individuals, and it is
important to correctly identify these groups since the efficiency of resulting
estimators can be improved significantly by pooling the information of
individuals within each group. However, the currently assumed parametric and
semiparametric relationship between the response and predictors may be
misspecified, which leads to a wrong grouping result, and the nonparametric
approach hence can be considered to avoid such mistakes. Moreover, the response
may depend on predictors in different ways at various quantile levels, and the
corresponding grouping structure may also vary. To tackle these problems, this
paper proposes a nonparametric quantile regression method for homogeneity
pursuit, and a pairwise fused penalty is used to automatically select the
number of groups. The asymptotic properties are established, and an ADMM
algorithm is also developed. The finite sample performance is evaluated by
simulation experiments, and the usefulness of the proposed methodology is
further illustrated by an empirical example.
| stat |
Finding manoeuvre motifs in vehicle telematics | Driving behaviour has a great impact on road safety. A popular way of
analysing driving behaviour is to move the focus to the manoeuvres as they give
useful information about the driver who is performing them. In this paper, we
investigate a new way of identifying manoeuvres from vehicle telematics data,
through motif detection in time-series. We implement a modified version of the
Extended Motif Discovery (EMD) algorithm, a classical variable-length motif
detection algorithm for time-series and we applied it to the UAH-DriveSet, a
publicly available naturalistic driving dataset. After a systematic exploration
of the extracted motifs, we were able to conclude that the EMD algorithm was
not only capable of extracting simple manoeuvres such as accelerations, brakes
and curves, but also more complex manoeuvres, such as lane changes and
overtaking manoeuvres, which validates motif discovery as a worthwhile line for
future research.
| stat |
Returning Scientists and the Emergence of Chinese Science System | The Chinese approach to developing a world-class science system includes a
vigorous set of programmes to attract back Chinese researchers who have
overseas training and work experience. No analysis is available to show the
performance of these mobile researchers. This article attempts to close part of
this gap. Using a novel bibliometric approach, we estimate the stocks of
overseas Chinese and returnees from the perspective of their publication
activities, albeit with some limitations. We show that the share of overseas
Chinese scientists in the US is considerably larger than that in the EU. We
also show that Chinese returnees publish higher impact work, and continue to
publish more and at the international level than domestic counterparts.
Returnees not only tend to publish more, but they are instrumental in linking
China into the global network. Indeed, returnees actively co-publish with
researchers in their former host system, showing the importance of scientific
social capital. Future research will examine the impact of length of stay,
among other factors, on such impact and integration.
| stat |
The Cost of Privacy in Generalized Linear Models: Algorithms and Minimax
Lower Bounds | We propose differentially private algorithms for parameter estimation in both
low-dimensional and high-dimensional sparse generalized linear models (GLMs) by
constructing private versions of projected gradient descent. We show that the
proposed algorithms are nearly rate-optimal by characterizing their statistical
performance and establishing privacy-constrained minimax lower bounds for GLMs.
The lower bounds are obtained via a novel technique, which is based on Stein's
Lemma and generalizes the tracing attack technique for privacy-constrained
lower bounds. This lower bound argument can be of independent interest as it is
applicable to general parametric models. Simulated and real data experiments
are conducted to demonstrate the numerical performance of our algorithms.
| stat |
Calibrated Optimal Decision Making with Multiple Data Sources and
Limited Outcome | We consider the optimal decision-making problem in a primary sample of
interest with multiple auxiliary sources available. The outcome of interest is
limited in the sense that it is only observed in the primary sample. In
reality, such multiple data sources may belong to different populations and
thus cannot be combined directly. This paper proposes a novel calibrated
optimal decision rule (CODR) to address the limited outcome, by leveraging the
shared pattern in multiple data sources. Under a mild and testable assumption
that the conditional means of intermediate outcomes in different samples are
equal given baseline covariates and the treatment information, we can show that
the calibrated mean outcome of interest under the CODR is unbiased and more
efficient than using the primary sample solely. Extensive experiments on
simulated datasets demonstrate empirical validity and improvement of the
proposed CODR, followed by a real application on the MIMIC-III as the primary
sample with auxiliary data from eICU.
| stat |
Persistent Reductions in Regularized Loss Minimization for Variable
Selection | In the context of regularized loss minimization with polyhedral gauges, we
show that for a broad class of loss functions (possibly non-smooth and
non-convex) and under a simple geometric condition on the input data it is
possible to efficiently identify a subset of features which are guaranteed to
have zero coefficients in all optimal solutions in all problems with loss
functions from said class, before any iterative optimization has been performed
for the original problem. This procedure is standalone, takes only the data as
input, and does not require any calls to the loss function. Therefore, we term
this procedure as a persistent reduction for the aforementioned class of
regularized loss minimization problems. This reduction can be efficiently
implemented via an extreme ray identification subroutine applied to a
polyhedral cone formed from the datapoints. We employ an existing
output-sensitive algorithm for extreme ray identification which makes our
guarantee and algorithm applicable in ultra-high dimensional problems.
| stat |
A Bayesian Framework for Persistent Homology | Persistence diagrams offer a way to summarize topological and geometric
properties latent in datasets. While several methods have been developed that
utilize persistence diagrams in statistical inference, a full Bayesian
treatment remains absent. This paper, relying on the theory of point processes,
presents a Bayesian framework for inference with persistence diagrams relying
on a substitution likelihood argument. In essence, we model persistence
diagrams as Poisson point processes with prior intensities and compute
posterior intensities by adopting techniques from the theory of marked point
processes. We then propose a family of conjugate prior intensities via Gaussian
mixtures to obtain a closed form of the posterior intensity. Finally we
demonstrate the utility of this Bayesian framework with a classification
problem in materials science using Bayes factors.
| stat |
Characterization of Sine- Skewed von Mises Distribution | The von Mises distribution is one of the most important distribution in
statistics to deal with circular data. In this paper we will consider some
basic properties and characterizations of the sine skewed von Mises
distribution.
| stat |
Bayesian data fusion for unmeasured confounding | Bayesian causal inference offers a principled approach to policy evaluation
of proposed interventions on mediators or time-varying exposures. We outline a
general approach to the estimation of causal quantities for settings with
time-varying confounding, such as exposure-induced mediator-outcome
confounders. We further extend this approach to propose two Bayesian data
fusion (BDF) methods for unmeasured confounding. Using informative priors on
quantities relating to the confounding bias parameters, our methods incorporate
data from an external source where the confounder is measured in order to make
inferences about causal estimands in the main study population. We present
results from a simulation study comparing our data fusion methods to two common
frequentist correction methods for unmeasured confounding bias in the mediation
setting. We also demonstrate our method with an investigation of the role of
stage at cancer diagnosis in contributing to Black-White colorectal cancer
survival disparities.
| stat |
Measuring Sample Quality with Kernels | Approximate Markov chain Monte Carlo (MCMC) offers the promise of more rapid
sampling at the cost of more biased inference. Since standard MCMC diagnostics
fail to detect these biases, researchers have developed computable Stein
discrepancy measures that provably determine the convergence of a sample to its
target distribution. This approach was recently combined with the theory of
reproducing kernels to define a closed-form kernel Stein discrepancy (KSD)
computable by summing kernel evaluations across pairs of sample points. We
develop a theory of weak convergence for KSDs based on Stein's method,
demonstrate that commonly used KSDs fail to detect non-convergence even for
Gaussian targets, and show that kernels with slowly decaying tails provably
determine convergence for a large class of target distributions. The resulting
convergence-determining KSDs are suitable for comparing biased, exact, and
deterministic sample sequences and simpler to compute and parallelize than
alternative Stein discrepancies. We use our tools to compare biased samplers,
select sampler hyperparameters, and improve upon existing KSD approaches to
one-sample hypothesis testing and sample quality improvement.
| stat |
MARGIN: Uncovering Deep Neural Networks using Graph Signal Analysis | Interpretability has emerged as a crucial aspect of building trust in machine
learning systems, aimed at providing insights into the working of complex
neural networks that are otherwise opaque to a user. There are a plethora of
existing solutions addressing various aspects of interpretability ranging from
identifying prototypical samples in a dataset to explaining image predictions
or explaining mis-classifications. While all of these diverse techniques
address seemingly different aspects of interpretability, we hypothesize that a
large family of interepretability tasks are variants of the same central
problem which is identifying \emph{relative} change in a model's prediction.
This paper introduces MARGIN, a simple yet general approach to address a large
set of interpretability tasks MARGIN exploits ideas rooted in graph signal
analysis to determine influential nodes in a graph, which are defined as those
nodes that maximally describe a function defined on the graph. By carefully
defining task-specific graphs and functions, we demonstrate that MARGIN
outperforms existing approaches in a number of disparate interpretability
challenges.
| stat |
Hidden Markov Chains, Entropic Forward-Backward, and Part-Of-Speech
Tagging | The ability to take into account the characteristics - also called features -
of observations is essential in Natural Language Processing (NLP) problems.
Hidden Markov Chain (HMC) model associated with classic Forward-Backward
probabilities cannot handle arbitrary features like prefixes or suffixes of any
size, except with an independence condition. For twenty years, this default has
encouraged the development of other sequential models, starting with the
Maximum Entropy Markov Model (MEMM), which elegantly integrates arbitrary
features. More generally, it led to neglect HMC for NLP. In this paper, we show
that the problem is not due to HMC itself, but to the way its restoration
algorithms are computed. We present a new way of computing HMC based
restorations using original Entropic Forward and Entropic Backward (EFB)
probabilities. Our method allows taking into account features in the HMC
framework in the same way as in the MEMM framework. We illustrate the
efficiency of HMC using EFB in Part-Of-Speech Tagging, showing its superiority
over MEMM based restoration. We also specify, as a perspective, how HMCs with
EFB might appear as an alternative to Recurrent Neural Networks to treat
sequential data with a deep architecture.
| stat |
Personalized Activity Recognition with Deep Triplet Embeddings | A significant challenge for a supervised learning approach to inertial human
activity recognition is the heterogeneity of data between individual users,
resulting in very poor performance of impersonal algorithms for some subjects.
We present an approach to personalized activity recognition based on deep
embeddings derived from a fully convolutional neural network. We experiment
with both categorical cross entropy loss and triplet loss for training the
embedding, and describe a novel triplet loss function based on subject
triplets. We evaluate these methods on three publicly available inertial human
activity recognition data sets (MHEALTH, WISDM, and SPAR) comparing
classification accuracy, out-of-distribution activity detection, and embedding
generalization to new activities. The novel subject triplet loss provides the
best performance overall, and all personalized deep embeddings out-perform our
baseline personalized engineered feature embedding and an impersonal fully
convolutional neural network classifier.
| stat |
Computation and application of generalized linear mixed model
derivatives using lme4 | Maximum likelihood estimation of generalized linear mixed models(GLMMs) is
difficult due to marginalization of the random effects. Computing derivatives
of a fitted GLMM's likelihood (with respect to model parameters) is also
difficult, especially because the derivatives are not by-products of popular
estimation algorithms. In this paper, we describe GLMM derivatives along with a
quadrature method to efficiently compute them, focusing on lme4 models with a
single clustering variable. We describe how psychometric results related to IRT
are helpful for obtaining these derivatives, as well as for verifying the
derivatives' accuracies. After describing the derivative computation methods,
we illustrate the many possible uses of these derivatives, including robust
standard errors, score tests of fixed effect parameters, and likelihood ratio
tests of non-nested models. The derivative computation methods and applications
described in the paper are all available in easily-obtained R packages.
| stat |
Optimizing the Latent Space of Generative Networks | Generative Adversarial Networks (GANs) have achieved remarkable results in
the task of generating realistic natural images. In most successful
applications, GAN models share two common aspects: solving a challenging saddle
point optimization problem, interpreted as an adversarial game between a
generator and a discriminator functions; and parameterizing the generator and
the discriminator as deep convolutional neural networks. The goal of this paper
is to disentangle the contribution of these two factors to the success of GANs.
In particular, we introduce Generative Latent Optimization (GLO), a framework
to train deep convolutional generators using simple reconstruction losses.
Throughout a variety of experiments, we show that GLO enjoys many of the
desirable properties of GANs: synthesizing visually-appealing samples,
interpolating meaningfully between samples, and performing linear arithmetic
with noise vectors; all of this without the adversarial optimization scheme.
| stat |
Supervised classification via minimax probabilistic transformations | Conventional techniques for supervised classification constrain the
classification rules considered and use surrogate losses for classification 0-1
loss. Favored families of classification rules are those that enjoy parametric
representations suitable for surrogate loss minimization, and low complexity
properties suitable for overfitting control. This paper presents classification
techniques based on robust risk minimization (RRM) that we call linear
probabilistic classifiers (LPCs). The proposed techniques consider
unconstrained classification rules, optimize the classification 0-1 loss, and
provide performance bounds during learning. LPCs enable efficient learning by
using linear optimization, and avoid overffiting by using RRM over polyhedral
uncertainty sets of distributions. We also provide finite-sample generalization
bounds for LPCs and show their competitive performance with state-of-the-art
techniques using benchmark datasets.
| stat |
On Comparison Of Experts | A policy maker faces a sequence of unknown outcomes. At each stage two
(self-proclaimed) experts provide probabilistic forecasts on the outcome in the
next stage. A comparison test is a protocol for the policy maker to
(eventually) decide which of the two experts is better informed. The protocol
takes as input the sequence of pairs of forecasts and actual realizations and
(weakly) ranks the two experts. We propose two natural properties that such a
comparison test must adhere to and show that these essentially uniquely
determine the comparison test. This test is a function of the derivative of the
induced pair of measures at the realization.
| stat |
HINT: A Hierarchical Independent Component Analysis Toolbox for
Investigating Brain Functional Networks using Neuroimaging Data | Independent component analysis (ICA) is a popular tool for investigating
brain organization in neuroscience research. In fMRI studies, an important goal
is to study how brain networks are modulated by subjects' clinical and
demographic variables. Existing ICA methods and toolboxes don't incorporate
subjects' covariates effects in ICA estimation of brain networks, which
potentially leads to loss in accuracy and statistical power in detecting brain
network differences between subjects' groups. We introduce a Matlab toolbox,
HINT (Hierarchical INdependent component analysis Toolbox), that provides a
hierarchical covariate-adjusted ICA (hc-ICA) for modeling and testing covariate
effects and generates model-based estimates of brain networks on both the
population- and individual-level. HINT provides a user-friendly Matlab GUI that
allows users to easily load images, specify covariate effects, monitor model
estimation via an EM algorithm, specify hypothesis tests, and visualize
results. HINT also has a command line interface which allows users to
conveniently run and reproduce the analysis with a script. HINT implements a
new multi-level probabilistic ICA model for group ICA. It provides a
statistically principled ICA modeling framework for investigating covariate
effects on brain networks. HINT can also generate and visualize model-based
network estimates for user-specified subject groups, which greatly facilitates
group comparisons.
| stat |
On the Insufficiency of the Large Margins Theory in Explaining the
Performance of Ensemble Methods | Boosting and other ensemble methods combine a large number of weak
classifiers through weighted voting to produce stronger predictive models. To
explain the successful performance of boosting algorithms, Schapire et al.
(1998) showed that AdaBoost is especially effective at increasing the margins
of the training data. Schapire et al. (1998) also developed an upper bound on
the generalization error of any ensemble based on the margins of the training
data, from which it was concluded that larger margins should lead to lower
generalization error, everything else being equal (sometimes referred to as the
``large margins theory''). Tighter bounds have been derived and have reinforced
the large margins theory hypothesis. For instance, Wang et al. (2011) suggest
that specific margin instances, such as the equilibrium margin, can better
summarize the margins distribution. These results have led many researchers to
consider direct optimization of the margins to improve ensemble generalization
error with mixed results. We show that the large margins theory is not
sufficient for explaining the performance of voting classifiers. We do this by
illustrating how it is possible to improve upon the margin distribution of an
ensemble solution, while keeping the complexity fixed, yet not improve the test
set performance.
| stat |
Bayesian Multi-scale Modeling of Factor Matrix without using Partition
Tree | The multi-scale factor models are particularly appealing for analyzing
matrix- or tensor-valued data, due to their adaptiveness to local geometry and
intuitive interpretation. However, the reliance on the binary tree for
recursive partitioning creates high complexity in the parameter space, making
it extremely challenging to quantify its uncertainty. In this article, we
discover an alternative way to generate multi-scale matrix using simple matrix
operation: starting from a random matrix with each column having two unique
values, its Cholesky whitening transform obeys a recursive partitioning
structure. This allows us to consider a generative distribution with large
prior support on common multi-scale factor models, and efficient posterior
computation via Hamiltonian Monte Carlo. We demonstrate its potential in a
multi-scale factor model to find broader regions of interest for human brain
connectivity.
| stat |
Principal Component Analysis on the Philippine Health Data | This study was conducted to determine the structures of a set of n correlated
variables and creates a new set of uncorrelated indices which are the
underlying components of the Philippine health data.The data utilized in this
study was the 2009 Philippine Health Data which was made available by National
Statistical Coordination Board(NSCB) in its 2009 publication.The publication
contains the health data of 81 provinces of the Philippines consisting of ten
system-related determinants which was considered as the variables in this
study. From the ten health system-related determinants, it was found out that
there are three significant underlying components that could summarize the
Philippine health data. The first component was named as importance of safe
water supply and emphasis on child heat while the second and third component
were named as importance of Barangay Health Stations, government health workers
and emphasis on pregnant women's health and emphasis on women's health,
respectively. These three components jointly account for a total of 73.01% of
the total variance explained by the component.
| stat |
Subsets and Splits
No saved queries yet
Save your SQL queries to embed, download, and access them later. Queries will appear here once saved.