title
stringlengths 6
244
| abstract
stringlengths 19
5.09k
| label
stringclasses 10
values |
|---|---|---|
Deep Quantile Aggregation | Conditional quantile estimation is a key statistical learning challenge
motivated by the need to quantify uncertainty in predictions or to model a
diverse population without being overly reductive. As such, many models have
been developed for this problem. Adopting a meta viewpoint, we propose a
general framework (inspired by neural network optimization) for aggregating any
number of conditional quantile models in order to boost predictive accuracy. We
consider weighted ensembling strategies of increasing flexibility where the
weights may vary over individual models, quantile levels, and feature values.
An appeal of our approach is its portability: we ensure that estimated
quantiles at adjacent levels do not cross by applying simple transformations
through which gradients can be backpropagated, and this allows us to leverage
the modern deep learning toolkit for building quantile ensembles. Our
experiments confirm that ensembling can lead to big gains in accuracy, even
when the constituent models are themselves powerful and flexible.
| stat |
Universal Consistency of Decision Trees in High Dimensions | This paper shows that decision trees constructed with Classification and
Regression Trees (CART) methodology are universally consistent in an additive
model context, even when the number of predictor variables scales exponentially
with the sample size, under certain $1$-norm sparsity constraints. The
consistency is universal in the sense that there are no a priori assumptions on
the distribution of the predictor variables. Amazingly, this adaptivity to
(approximate or exact) sparsity is achieved with a single tree, as opposed to
what might be expected for an ensemble. Finally, we show that these qualitative
properties of individual trees are inherited by Breiman's random forests.
Another surprise is that consistency holds even when the "mtry" tuning
parameter vanishes as a fraction of the number of predictor variables, thus
speeding up computation of the forest. A key step in the analysis is the
establishment of an oracle inequality, which precisely characterizes the
goodness-of-fit and complexity tradeoff for a misspecified model.
| stat |
A Regionalisation Approach for Rainfall based on Extremal Dependence | To mitigate the risk posed by extreme rainfall events, we require statistical
models that reliably capture extremes in continuous space with dependence.
However, assuming a stationary dependence structure in such models is often
erroneous, particularly over large geographical domains. Furthermore, there are
limitations on the ability to fit existing models, such as max-stable
processes, to a large number of locations. To address these modelling
challenges, we present a regionalisation method that partitions stations into
regions of similar extremal dependence using clustering. To demonstrate our
regionalisation approach, we consider a study region of Australia and discuss
the results with respect to known climate and topographic features. To
visualise and evaluate the effectiveness of the partitioning, we fit max-stable
models to each of the regions. This work serves as a prelude to how one might
consider undertaking a project where spatial dependence is non-stationary and
is modelled on a large geographical scale.
| stat |
Forecasting the Intra-Day Spread Densities of Electricity Prices | Intra-day price spreads are of interest to electricity traders, storage and
electric vehicle operators. This paper formulates dynamic density functions,
based upon skewed-t and similar representations, to model and forecast the
German electricity price spreads between different hours of the day, as
revealed in the day-ahead auctions. The four specifications of the density
functions are dynamic and conditional upon exogenous drivers, thereby
permitting the location, scale and shape parameters of the densities to respond
hourly to such factors as weather and demand forecasts. The best fitting and
forecasting specifications for each spread are selected based on the Pinball
Loss function, following the closed-form analytical solutions of the cumulative
distribution functions.
| stat |
Robust spectral clustering using LASSO regularization | Cluster structure detection is a fundamental task for the analysis of graphs,
in order to understand and to visualize their functional characteristics. Among
the different cluster structure detection methods, spectral clustering is
currently one of the most widely used due to its speed and simplicity. Yet,
there are few theoretical guarantee to recover the underlying partitions of the
graph for general models. This paper therefore presents a variant of spectral
clustering, called 1-spectral clustering, performed on a new random model
closely related to stochastic block model. Its goal is to promote a sparse
eigenbasis solution of a 1 minimization problem revealing the natural structure
of the graph. The effectiveness and the robustness to small noise perturbations
of our technique is confirmed through a collection of simulated and real data
examples.
| stat |
Modelling Complex Survey Data Using R, SAS, SPSS and Stata: A Comparison
Using CLSA Datasets | The R software has become popular among researchers due to its flexibility
and open-source nature. However, researchers in the fields of public health and
epidemiological studies are more customary to commercial statistical softwares
such as SAS, SPSS and Stata. This paper provides a comprehensive comparison on
analysis of health survey data using the R survey package, SAS, SPSS and Stata.
We describe detailed R codes and procedures for other software packages on
commonly encountered statistical analyses, such as estimation of population
means and regression analysis, using datasets from the Canadian Longitudinal
Study on Aging (CLSA). It is hoped that the paper stimulates interest among
health science researchers to carry data analysis using R and also serves as a
cookbook for statistical analysis using different software packages.
| stat |
Bayesian Calibration of Computer Models with Informative Failures | There are many practical difficulties in the calibration of computer models
to experimental data. One such complication is the fact that certain
combinations of the calibration inputs can cause the code to output data
lacking fundamental properties, or even to produce no output at all. In many
cases the researchers want or need to exclude the possibility of these
"failures" within their analyses. We propose a Bayesian (meta-)model in which
the posterior distribution for the calibration parameters naturally excludes
regions of the input space corresponding to failed runs. That is, we define a
statistical selection model to rigorously couple the disjoint problems of
binary classification and computer model calibration. We demonstrate our
methodology using data from a carbon capture experiment in which the numerics
of the computational fluid dynamics are prone to instability.
| stat |
Modeling of Lifeline Infrastructure Restoration Using Empirical
Quantitative Data | Disaster recovery is widely regarded as the least understood phase of the
disaster cycle. In particular, the literature around lifeline infrastructure
restoration modeling frequently mentions the lack of empirical quantitative
data available. Despite limitations, there is a growing body of research on
modeling lifeline infrastructure restoration, often developed using empirical
quantitative data. This study reviews this body of literature and identifies
the data collection and usage patterns present across modeling approaches to
inform future efforts using empirical quantitative data. We classify the
modeling approaches into simulation, optimization, and statistical modeling.
The number of publications in this domain has increased over time with the most
rapid growth of statistical modeling. Electricity infrastructure restoration is
most frequently modeled, followed by the restoration of multiple
infrastructures, water infrastructure, and transportation infrastructure.
Interdependency between multiple infrastructures is increasingly considered in
recent literature. Researchers gather the data from a variety of sources,
including collaborations with utility companies, national databases, and
post-event damage and restoration reports. This study provides discussion and
recommendations around data usage practices within the lifeline restoration
modeling field. Following the recommendations would facilitate the development
of a community of practice around restoration modeling and provide greater
opportunities for future data sharing.
| stat |
Online Debiasing for Adaptively Collected High-dimensional Data with
Applications to Time Series Analysis | Adaptive collection of data is commonplace in applications throughout science
and engineering. From the point of view of statistical inference however,
adaptive data collection induces memory and correlation in the samples, and
poses significant challenge. We consider the high-dimensional linear
regression, where the samples are collected adaptively, and the sample size $n$
can be smaller than $p$, the number of covariates. In this setting, there are
two distinct sources of bias: the first due to regularization imposed for
consistent estimation, e.g. using the LASSO, and the second due to adaptivity
in collecting the samples. We propose "online debiasing", a general procedure
for estimators such as the LASSO, which addresses both sources of bias. In two
concrete contexts $(i)$ time series analysis and $(ii)$ batched data
collection, we demonstrate that online debiasing optimally debiases the LASSO
estimate when the underlying parameter $\theta_0$ has sparsity of order
$o(\sqrt{n}/\log p)$. In this regime, the debiased estimator can be used to
compute $p$-values and confidence intervals of optimal size.
| stat |
A Bayesian Markov model with P\'olya-Gamma sampling for estimating
individual behavior transition probabilities from accelerometer
classifications | The use of accelerometers in wildlife tracking provides a fine-scale data
source for understanding animal behavior and decision-making. Current methods
in movement ecology focus on behavior as a driver of movement mechanisms. Our
Markov model is a flexible and efficient method for inference related to
effects on behavior that considers dependence between current and past
behaviors. We applied this model to behavior data from six greater
white-fronted geese (Anser albifrons frontalis) during spring migration in
mid-continent North America and considered likely drivers of behavior,
including habitat, weather and time of day effects. We modeled the transitions
between flying, feeding, stationary and walking behavior states using a
first-order Bayesian Markov model. We introduced P\'olya-Gamma latent variables
for automatic sampling of the covariate coefficients from the posterior
distribution and we calculated the odds ratios from the posterior samples. Our
model provides a unifying framework for including both acceleration and Global
Positioning System data. We found significant differences in behavioral
transition rates among habitat types, diurnal behavior and behavioral changes
due to weather. Our model provides straightforward inference of behavioral time
allocation across used habitats, which is not amenable in activity budget or
resource selection frameworks.
| stat |
Statistical properties of sketching algorithms | Sketching is a probabilistic data compression technique that has been largely
developed in the computer science community. Numerical operations on big
datasets can be intolerably slow; sketching algorithms address this issue by
generating a smaller surrogate dataset. Typically, inference proceeds on the
compressed dataset. Sketching algorithms generally use random projections to
compress the original dataset and this stochastic generation process makes them
amenable to statistical analysis. We argue that the sketched data can be
modelled as a random sample, thus placing this family of data compression
methods firmly within an inferential framework. In particular, we focus on the
Gaussian, Hadamard and Clarkson-Woodruff sketches, and their use in single pass
sketching algorithms for linear regression with huge $n$. We explore the
statistical properties of sketched regression algorithms and derive new
distributional results for a large class of sketched estimators. A key result
is a conditional central limit theorem for data oblivious sketches. An
important finding is that the best choice of sketching algorithm in terms of
mean square error is related to the signal to noise ratio in the source
dataset. Finally, we demonstrate the theory and the limits of its applicability
on two real datasets.
| stat |
A Parsimonious Personalized Dose Finding Model via Dimension Reduction | Learning an individualized dose rule in personalized medicine is a
challenging statistical problem. Existing methods for estimating the optimal
individualized dose rule often suffer from the curse of dimensionality,
especially when the dose rule is learned nonparametrically using machine
learning approaches. To tackle this problem, we propose a dimension reduction
framework that effectively reduces the estimation of dose rule in a
lower-dimensional subspace of the covariates, leading to a more parsimonious
model. To achieve this, the proposed methods exploit that the subspace is
spanned by a few linear combinations of the covariates, which can be estimated
efficiently by using an orthogonality constrained optimization approach. Using
this framework, we proposed a direct estimation of the value function under any
given suggested dose rule with dimension reduction. This does not require an
inverse probability of the propensity score, which distinguishes us from the
popular outcome weighted learning framework. We further propose two approaches:
a direct learning approach that yields the dose rule as commonly desired in
personalized medicine, and a pseudo-direct learning approach that focuses more
on estimating the dimensionality-reduced subspace. Under mild regularity
assumptions, the asymptotical normality of the proposed subspace estimators is
established respectively. The consistency and convergence rate for the
estimated optimal dose rule are also derived. For both approaches, we formulate
an effectively numerical optimization problem as solving solutions on the
Stiefel manifold. The performance of the proposed approaches is evaluated
through simulation studies and a warfarin pharmacogenetic dataset.
| stat |
Active Manifolds: A non-linear analogue to Active Subspaces | We present an approach to analyze $C^1(\mathbb{R}^m)$ functions that
addresses limitations present in the Active Subspaces (AS) method of
Constantine et al.(2015; 2014). Under appropriate hypotheses, our Active
Manifolds (AM) method identifies a 1-D curve in the domain (the active
manifold) on which nearly all values of the unknown function are attained, and
which can be exploited for approximation or analysis, especially when $m$ is
large (high-dimensional input space). We provide theorems justifying our AM
technique and an algorithm permitting functional approximation and sensitivity
analysis. Using accessible, low-dimensional functions as initial examples, we
show AM reduces approximation error by an order of magnitude compared to AS, at
the expense of more computation. Following this, we revisit the sensitivity
analysis by Glaws et al. (2017), who apply AS to analyze a magnetohydrodynamic
power generator model, and compare the performance of AM on the same data. Our
analysis provides detailed information not captured by AS, exhibiting the
influence of each parameter individually along an active manifold. Overall, AM
represents a novel technique for analyzing functional models with benefits
including: reducing $m$-dimensional analysis to a 1-D analogue, permitting more
accurate regression than AS (at more computational expense), enabling more
informative sensitivity analysis, and granting accessible visualizations(2-D
plots) of parameter sensitivity along the AM.
| stat |
How to Measure Your App: A Couple of Pitfalls and Remedies in Measuring
App Performance in Online Controlled Experiments | Effectively measuring, understanding, and improving mobile app performance is
of paramount importance for mobile app developers. Across the mobile Internet
landscape, companies run online controlled experiments (A/B tests) with
thousands of performance metrics in order to understand how app performance
causally impacts user retention and to guard against service or app regressions
that degrade user experiences. To capture certain characteristics particular to
performance metrics, such as enormous observation volume and high skewness in
distribution, an industry-standard practice is to construct a performance
metric as a quantile over all performance events in control or treatment
buckets in A/B tests. In our experience with thousands of A/B tests provided by
Snap, we have discovered some pitfalls in this industry-standard way of
calculating performance metrics that can lead to unexplained movements in
performance metrics and unexpected misalignment with user engagement metrics.
In this paper, we discuss two major pitfalls in this industry-standard practice
of measuring performance for mobile apps. One arises from strong heterogeneity
in both mobile devices and user engagement, and the other arises from
self-selection bias caused by post-treatment user engagement changes. To remedy
these two pitfalls, we introduce several scalable methods including user-level
performance metric calculation and imputation and matching for missing metric
values. We have extensively evaluated these methods on both simulation data and
real A/B tests, and have deployed them into Snap's in-house experimentation
platform.
| stat |
Task Agnostic Continual Learning Using Online Variational Bayes with
Fixed-Point Updates | Background: Catastrophic forgetting is the notorious vulnerability of neural
networks to the changes in the data distribution during learning. This
phenomenon has long been considered a major obstacle for using learning agents
in realistic continual learning settings. A large body of continual learning
research assumes that task boundaries are known during training. However, only
a few works consider scenarios in which task boundaries are unknown or not well
defined -- task agnostic scenarios. The optimal Bayesian solution for this
requires an intractable online Bayes update to the weights posterior.
Contributions: We aim to approximate the online Bayes update as accurately as
possible. To do so, we derive novel fixed-point equations for the online
variational Bayes optimization problem, for multivariate Gaussian parametric
distributions. By iterating the posterior through these fixed-point equations,
we obtain an algorithm (FOO-VB) for continual learning which can handle
non-stationary data distribution using a fixed architecture and without using
external memory (i.e. without access to previous data). We demonstrate that our
method (FOO-VB) outperforms existing methods in task agnostic scenarios. FOO-VB
Pytorch implementation will be available online.
| stat |
Generalized inferential models for censored data | Inferential challenges that arise when data are censored have been
extensively studied under the classical frameworks. In this paper, we provide
an alternative generalized inferential model approach whose output is a
data-dependent plausibility function. This construction is driven by an
association between the distribution of the relative likelihood function at the
interest parameter and an unobserved auxiliary variable. The plausibility
function emerges from the distribution of a suitably calibrated random set
designed to predict that unobserved auxiliary variable. The evaluation of this
plausibility function requires a novel use of the classical Kaplan--Meier
estimator to estimate the censoring rather than the event distribution. We
prove that the proposed method provides valid inference, at least
approximately, and our real- and simulated-data examples demonstrate its
superior performance compared to existing methods.
| stat |
Approximate Message Passing with Spectral Initialization for Generalized
Linear Models | We consider the problem of estimating a signal from measurements obtained via
a generalized linear model. We focus on estimators based on approximate message
passing (AMP), a family of iterative algorithms with many appealing features:
the performance of AMP in the high-dimensional limit can be succinctly
characterized under suitable model assumptions; AMP can also be tailored to the
empirical distribution of the signal entries, and for a wide class of
estimation problems, AMP is conjectured to be optimal among all polynomial-time
algorithms.
However, a major issue of AMP is that in many models (such as phase
retrieval), it requires an initialization correlated with the ground-truth
signal and independent from the measurement matrix. Assuming that such an
initialization is available is typically not realistic. In this paper, we solve
this problem by proposing an AMP algorithm initialized with a spectral
estimator. With such an initialization, the standard AMP analysis fails since
the spectral estimator depends in a complicated way on the design matrix. Our
main contribution is a rigorous characterization of the performance of AMP with
spectral initialization in the high-dimensional limit. The key technical idea
is to define and analyze a two-phase artificial AMP algorithm that first
produces the spectral estimator, and then closely approximates the iterates of
the true AMP. We also provide numerical results that demonstrate the validity
of the proposed approach.
| stat |
Doubts on the efficacy of outliers correction methods | While the utilisation of different methods of outliers correction has been
shown to counteract the inferential error produced by the presence of
contaminating data not belonging to the studied population; the effects
produced by their utilisation when samples do not contain contaminating
outliers are less clear. Here a simulation approach shows that the most popular
methods of outliers correction (2 Sigma, 3 Sigma, MAD, IQR, Grubbs and
winsorizing) worsen the inferential evaluation of the studied population in
this condition, in particular producing an inflation of Type I error and
increasing the error committed in estimating the population mean and STD. We
show that those methods that have the highest efficacy in counteract the
inflation of Type I and Type II errors in the presence of contaminating
outliers also produce the stronger increase of false positive results in their
absence, suggesting that the systematic utilisation of methods for outliers
correction risk to produce more harmful than beneficial effect on statistical
inference. We finally propose that the safest way to deal with the presence of
outliers for statistical comparisons is the utilisation of non-parametric tests
| stat |
Shape and Time Distortion Loss for Training Deep Time Series Forecasting
Models | This paper addresses the problem of time series forecasting for
non-stationary signals and multiple future steps prediction. To handle this
challenging task, we introduce DILATE (DIstortion Loss including shApe and
TimE), a new objective function for training deep neural networks. DILATE aims
at accurately predicting sudden changes, and explicitly incorporates two terms
supporting precise shape and temporal change detection. We introduce a
differentiable loss function suitable for training deep neural nets, and
provide a custom back-prop implementation for speeding up optimization. We also
introduce a variant of DILATE, which provides a smooth generalization of
temporally-constrained Dynamic Time Warping (DTW). Experiments carried out on
various non-stationary datasets reveal the very good behaviour of DILATE
compared to models trained with the standard Mean Squared Error (MSE) loss
function, and also to DTW and variants. DILATE is also agnostic to the choice
of the model, and we highlight its benefit for training fully connected
networks as well as specialized recurrent architectures, showing its capacity
to improve over state-of-the-art trajectory forecasting approaches.
| stat |
Modeling treatment events in disease progression | Ability to quantify and predict progression of a disease is fundamental for
selecting an appropriate treatment. Many clinical metrics cannot be acquired
frequently either because of their cost (e.g. MRI, gait analysis) or because
they are inconvenient or harmful to a patient (e.g. biopsy, x-ray). In such
scenarios, in order to estimate individual trajectories of disease progression,
it is advantageous to leverage similarities between patients, i.e. the
covariance of trajectories, and find a latent representation of progression.
Most of existing methods for estimating trajectories do not account for events
in-between observations, what dramatically decreases their adequacy for
clinical practice. In this study, we develop a machine learning framework named
Coordinatewise-Soft-Impute (CSI) for analyzing disease progression from sparse
observations in the presence of confounding events. CSI is guaranteed to
converge to the global minimum of the corresponding optimization problem.
Experimental results also demonstrates the effectiveness of CSI using both
simulated and real dataset.
| stat |
Fast and Flexible Multi-Task Classification Using Conditional Neural
Adaptive Processes | The goal of this paper is to design image classification systems that, after
an initial multi-task training phase, can automatically adapt to new tasks
encountered at test time. We introduce a conditional neural process based
approach to the multi-task classification setting for this purpose, and
establish connections to the meta-learning and few-shot learning literature.
The resulting approach, called CNAPs, comprises a classifier whose parameters
are modulated by an adaptation network that takes the current task's dataset as
input. We demonstrate that CNAPs achieves state-of-the-art results on the
challenging Meta-Dataset benchmark indicating high-quality transfer-learning.
We show that the approach is robust, avoiding both over-fitting in low-shot
regimes and under-fitting in high-shot regimes. Timing experiments reveal that
CNAPs is computationally efficient at test-time as it does not involve gradient
based adaptation. Finally, we show that trained models are immediately
deployable to continual learning and active learning where they can outperform
existing approaches that do not leverage transfer learning.
| stat |
Multi-class Gaussian Process Classification with Noisy Inputs | It is a common practice in the machine learning community to assume that the
observed data are noise-free in the input attributes. Nevertheless, scenarios
with input noise are common in real problems, as measurements are never
perfectly accurate. If this input noise is not taken into account, a supervised
machine learning method is expected to perform sub-optimally. In this paper, we
focus on multi-class classification problems and use Gaussian processes (GPs)
as the underlying classifier. Motivated by a data set coming from the
astrophysics domain, we hypothesize that the observed data may contain noise in
the inputs. Therefore, we devise several multi-class GP classifiers that can
account for input noise. Such classifiers can be efficiently trained using
variational inference to approximate the posterior distribution of the latent
variables of the model. Moreover, in some situations, the amount of noise can
be known before-hand. If this is the case, it can be readily introduced in the
proposed methods. This prior information is expected to lead to better
performance results. We have evaluated the proposed methods by carrying out
several experiments, involving synthetic and real data. These include several
data sets from the UCI repository, the MNIST data set and a data set coming
from astrophysics. The results obtained show that, although the classification
error is similar across methods, the predictive distribution of the proposed
methods is better, in terms of the test log-likelihood, than the predictive
distribution of a classifier based on GPs that ignores input noise.
| stat |
Credit Crunch: The Role of Household Lending Capacity in the Dutch
Housing Boom and Bust 1995-2018 | What causes house prices to rise and fall? Economists identify household
access to credit as a crucial factor. "Loan-to-Value" and "Debt-to-GDP" ratios
are the standard measures for credit access. However, these measures fail to
explain the depth of the Dutch housing bust after the 2009 Financial Crisis.
This work is the first to model household lending capacity based on the
formulas that Dutch banks use in the mortgage application process. We compare
the ability of regression models to forecast housing prices when different
measures of credit access are utilised. We show that our measure of household
lending capacity is a forward-looking, highly predictive variable that
outperforms `Loan-to-Value' and debt ratios in forecasting the Dutch crisis.
Sharp declines in lending capacity foreshadow the market deceleration.
| stat |
The ASCCR Frame for Learning Essential Collaboration Skills | Statistics and data science are especially collaborative disciplines that
typically require practitioners to interact with many different people or
groups. Consequently, interdisciplinary collaboration skills are part of the
personal and professional skills essential for success as an applied
statistician or data scientist. These skills are learnable and teachable, and
learning and improving collaboration skills provides a way to enhance one's
practice of statistics and data science. To help individuals learn these skills
and organizations to teach them, we have developed a framework covering five
essential components of statistical collaboration: Attitude, Structure,
Content, Communication, and Relationship. We call this the ASCCR Frame. This
framework can be incorporated into formal training programs in the classroom or
on the job and can also be used by individuals through self-study. We show how
this framework can be applied specifically to statisticians and data scientists
to improve their collaboration skills and their interdisciplinary impact. We
believe that the ASCCR Frame can help organize and stimulate research and
teaching in interdisciplinary collaboration and call on individuals and
organizations to begin generating evidence regarding its effectiveness.
| stat |
Sliced Gromov-Wasserstein | Recently used in various machine learning contexts, the Gromov-Wasserstein
distance (GW) allows for comparing distributions whose supports do not
necessarily lie in the same metric space. However, this Optimal Transport (OT)
distance requires solving a complex non convex quadratic program which is most
of the time very costly both in time and memory. Contrary to GW, the
Wasserstein distance (W) enjoys several properties ({\em e.g.} duality) that
permit large scale optimization. Among those, the solution of W on the real
line, that only requires sorting discrete samples in 1D, allows defining the
Sliced Wasserstein (SW) distance. This paper proposes a new divergence based on
GW akin to SW. We first derive a closed form for GW when dealing with 1D
distributions, based on a new result for the related quadratic assignment
problem. We then define a novel OT discrepancy that can deal with large scale
distributions via a slicing approach and we show how it relates to the GW
distance while being $O(n\log(n))$ to compute. We illustrate the behavior of
this so called Sliced Gromov-Wasserstein (SGW) discrepancy in experiments where
we demonstrate its ability to tackle similar problems as GW while being several
order of magnitudes faster to compute.
| stat |
Consensus Monte Carlo for Random Subsets using Shared Anchors | We present a consensus Monte Carlo algorithm that scales existing Bayesian
nonparametric models for clustering and feature allocation to big data. The
algorithm is valid for any prior on random subsets such as partitions and
latent feature allocation, under essentially any sampling model. Motivated by
three case studies, we focus on clustering induced by a Dirichlet process
mixture sampling model, inference under an Indian buffet process prior with a
binomial sampling model, and with a categorical sampling model. We assess the
proposed algorithm with simulation studies and show results for inference with
three datasets: an MNIST image dataset, a dataset of pancreatic cancer
mutations, and a large set of electronic health records (EHR). Supplementary
materials for this article are available online.
| stat |
Time-to-Event Prediction with Neural Networks and Cox Regression | New methods for time-to-event prediction are proposed by extending the Cox
proportional hazards model with neural networks. Building on methodology from
nested case-control studies, we propose a loss function that scales well to
large data sets, and enables fitting of both proportional and non-proportional
extensions of the Cox model. Through simulation studies, the proposed loss
function is verified to be a good approximation for the Cox partial
log-likelihood. The proposed methodology is compared to existing methodologies
on real-world data sets, and is found to be highly competitive, typically
yielding the best performance in terms of Brier score and binomial
log-likelihood. A python package for the proposed methods is available at
https://github.com/havakv/pycox.
| stat |
Functional linear models for interval-valued data | Aggregation of large databases in a specific format is a frequently used
process to make the data easily manageable. Interval-valued data is one of the
data types that is generated by such an aggregation process. Using traditional
methods to analyze interval-valued data results in loss of information, and
thus, several interval-valued data models have been proposed to gather reliable
information from such data types. On the other hand, recent technological
developments have led to high dimensional and complex data in many application
areas, which may not be analyzed by traditional techniques. Functional data
analysis is one of the most commonly used techniques to analyze such complex
datasets. While the functional extensions of much traditional statistical
techniques are available, the functional form of the interval-valued data has
not been studied well. This paper introduces the functional forms of some
well-known regression models that take interval-valued data. The proposed
methods are based on the function-on-function regression model, where both the
response and predictor/s are functional. Through several Monte Carlo
simulations and empirical data analysis, the finite sample performance of the
proposed methods is evaluated and compared with the state-of-the-art.
| stat |
Diagnostics and Visualization of Point Process Models for Event Times on
a Social Network | Point process models have been used to analyze interaction event times on a
social network, in the hope to provides valuable insights for social science
research. However, the diagnostics and visualization of the modeling results
from such an analysis have received limited discussion in the literature. In
this paper, we develop a systematic set of diagnostic tools and visualizations
for point process models fitted to data from a network setting. We analyze the
residual process and Pearson residual on the network by inspecting their
structure and clustering structure. Equipped with these tools, we can validate
whether a model adequately captures the temporal and/or network structures in
the observed data. The utility of our approach is demonstrated using simulation
studies and point process models applied to a study of animal social
interactions.
| stat |
On Matching Pursuit and Coordinate Descent | Two popular examples of first-order optimization methods over linear spaces
are coordinate descent and matching pursuit algorithms, with their randomized
variants. While the former targets the optimization by moving along
coordinates, the latter considers a generalized notion of directions.
Exploiting the connection between the two algorithms, we present a unified
analysis of both, providing affine invariant sublinear $\mathcal{O}(1/t)$ rates
on smooth objectives and linear convergence on strongly convex objectives. As a
byproduct of our affine invariant analysis of matching pursuit, our rates for
steepest coordinate descent are the tightest known. Furthermore, we show the
first accelerated convergence rate $\mathcal{O}(1/t^2)$ for matching pursuit
and steepest coordinate descent on convex objectives.
| stat |
Clustering measure-valued data with Wasserstein barycenters | In this work, learning schemes for measure-valued data are proposed, i.e.
data that their structure can be more efficiently represented as probability
measures instead of points on $\R^d$, employing the concept of probability
barycenters as defined with respect to the Wasserstein metric. Such type of
learning approaches are highly appreciated in many fields where the
observational/experimental error is significant (e.g. astronomy, biology,
remote sensing, etc.) or the data nature is more complex and the traditional
learning algorithms are not applicable or effective to treat them (e.g. network
data, interval data, high frequency records, matrix data, etc.). Under this
perspective, each observation is identified by an appropriate probability
measure and the proposed statistical learning schemes rely on discrimination
criteria that utilize the geometric structure of the space of probability
measures through core techniques from the optimal transport theory. The
discussed approaches are implemented in two real world applications: (a)
clustering eurozone countries according to their observed government bond yield
curves and (b) classifying the areas of a satellite image to certain land uses
categories which is a standard task in remote sensing. In both case studies the
results are particularly interesting and meaningful while the accuracy obtained
is high.
| stat |
Statistical Inference after Kernel Ridge Regression Imputation under
item nonresponse | Imputation is a popular technique for handling missing data. We consider a
nonparametric approach to imputation using the kernel ridge regression
technique and propose consistent variance estimation. The proposed variance
estimator is based on a linearization approach which employs the entropy method
to estimate the density ratio. The root-n consistency of the imputation
estimator is established when a Sobolev space is utilized in the kernel ridge
regression imputation, which enables us to develop the proposed variance
estimator. Synthetic data experiments are presented to confirm our theory.
| stat |
Transforming variables to central normality | Many real data sets contain numerical features (variables) whose distribution
is far from normal (gaussian). Instead, their distribution is often skewed. In
order to handle such data it is customary to preprocess the variables to make
them more normal. The Box-Cox and Yeo-Johnson transformations are well-known
tools for this. However, the standard maximum likelihood estimator of their
transformation parameter is highly sensitive to outliers, and will often try to
move outliers inward at the expense of the normality of the central part of the
data. We propose a modification of these transformations as well as an
estimator of the transformation parameter that is robust to outliers, so the
transformed data can be approximately normal in the center and a few outliers
may deviate from it. It compares favorably to existing techniques in an
extensive simulation study and on real data.
| stat |
Machine Learning and the Internet of Things Enable Steam Flood
Optimization for Improved Oil Production | Recently developed machine learning techniques, in association with the
Internet of Things (IoT) allow for the implementation of a method of increasing
oil production from heavy-oil wells. Steam flood injection, a widely used
enhanced oil recovery technique, uses thermal and gravitational potential to
mobilize and dilute heavy oil in situ to increase oil production. In contrast
to traditional steam flood simulations based on principles of classic physics,
we introduce here an approach using cutting-edge machine learning techniques
that have the potential to provide a better way to describe the performance of
steam flood. We propose a workflow to address a category of time-series data
that can be analyzed with supervised machine learning algorithms and IoT. We
demonstrate the effectiveness of the technique for forecasting oil production
in steam flood scenarios. Moreover, we build an optimization system that
recommends an optimal steam allocation plan, and show that it leads to a 3%
improvement in oil production. We develop a minimum viable product on a cloud
platform that can implement real-time data collection, transfer, and storage,
as well as the training and implementation of a cloud-based machine learning
model. This workflow also offers an applicable solution to other problems with
similar time-series data structures, like predictive maintenance.
| stat |
Learning Traffic Flow Dynamics using Random Fields | This paper presents a mesoscopic traffic flow model that explicitly describes
the spatio-temporal evolution of the probability distributions of vehicle
trajectories. The dynamics are represented by a sequence of factor graphs,
which enable learning of traffic dynamics from limited Lagrangian measurements
using an efficient message passing technique. The approach ensures that
estimated speeds and traffic densities are non-negative with probability one.
The estimation technique is tested using vehicle trajectory datasets generated
using an independent microscopic traffic simulator and is shown to efficiently
reproduce traffic conditions with probe vehicle penetration levels as little as
10\%. The proposed algorithm is also compared with state-of-the-art traffic
state estimation techniques developed for the same purpose and it is shown that
the proposed approach can outperform the state-of-the-art techniques in terms
reconstruction accuracy.
| stat |
Fast Graph Filters for Decentralized Subspace Projection | A number of inference problems with sensor networks involve projecting a
measured signal onto a given subspace. In existing decentralized approaches,
sensors communicate with their local neighbors to obtain a sequence of iterates
that asymptotically converges to the desired projection. In contrast, the
present paper develops methods that produce these projections in a finite and
approximately minimal number of iterations. Building upon tools from graph
signal processing, the problem is cast as the design of a graph filter which,
in turn, is reduced to the design of a suitable graph shift operator.
Exploiting the eigenstructure of the projection and shift matrices leads to an
objective whose minimization yields approximately minimum-order graph filters.
To cope with the fact that this problem is not convex, the present work
introduces a novel convex relaxation of the number of distinct eigenvalues of a
matrix based on the nuclear norm of a Kronecker difference. To tackle the case
where there exists no graph filter capable of implementing a certain subspace
projection with a given network topology, a second optimization criterion is
presented to approximate the desired projection while trading the number of
iterations for approximation error. Two algorithms are proposed to optimize the
aforementioned criteria based on the alternating-direction method of
multipliers. An exhaustive simulation study demonstrates that the obtained
filters can effectively obtain subspace projections markedly faster than
existing algorithms.
| stat |
Kernel-based parameter estimation of dynamical systems with unknown
observation functions | A low-dimensional dynamical system is observed in an experiment as a
high-dimensional signal; for example, a video of a chaotic pendulums system.
Assuming that we know the dynamical model up to some unknown parameters, can we
estimate the underlying system's parameters by measuring its time-evolution
only once? The key information for performing this estimation lies in the
temporal inter-dependencies between the signal and the model. We propose a
kernel-based score to compare these dependencies. Our score generalizes a
maximum likelihood estimator for a linear model to a general nonlinear setting
in an unknown feature space. We estimate the system's underlying parameters by
maximizing the proposed score. We demonstrate the accuracy and efficiency of
the method using two chaotic dynamical systems - the double pendulum and the
Lorenz '63 model.
| stat |
Introduction to Geodetic Time Series Analysis | This contribution is the chapter 2 of the book "geodetic time series
analysis" (10.1007/978-3-030-21718-1). The book is dedicated to the art of
fitting a trajectory model to those geodetic time series in order to extract
accurate geophysical information with realistic error bars in geodymanics and
environmental geodesy related studies. In the vast amount of the literature
published on this topic in the past 25 years, we are specifically interested in
parametric algorithms which are estimating both functional and stochastic
models using various Bayesian statistical tools (maximum likelihood, Monte
Carlo Markov chain, Kalman filter, least squares variance component estimation,
information criteria). This chapter will focus on how the parameters of the
trajectory model can be estimated. It is meant to give researchers new to this
topic an easy introduction to the theory with references to key books and
articles where more details can be found. In addition, we hope that it
refreshes some of the details for the more experienced readers. We pay special
attention to the modelling of the noise which has received much attention in
the literature in the last years and highlight some of the numerical aspects.
| stat |
visTree: Visualization of Subgroups for a Decision Tree | Decision trees are flexible prediction models which are constructed to
quantify outcome-covariate relationships and characterize relevant population
subgroups. However, the standard graphical representation of fitted decision
trees highlights individual split points, and hence is suboptimal for
visualizing defined subgroups. In this paper, we present a novel visual
representation of decision trees which shifts the primary focus to
characterizing subgroups, both in terms of their defining covariates and their
outcome distribution. We implement our method in the \texttt{visTree} package,
which builds on the toolkit and infrastructure provided by the
\texttt{partykit} package and enables the visualization to be applied to varied
decision trees. Individual functions are demonstrated using data from the Box
Lunch study [French et al., 2014], a randomized trial to evaluate the effect of
exposure to different lunch sizes on energy intake and body weight among
working adults.
| stat |
A review of problem- and team-based methods for teaching statistics in
Higher Education | The teaching of statistics in higher education in the UK is still largely
lecture-based. This is despite recommendations such as those given by the
American Statistical Association's GAISE report that more emphasis should be
placed on active learning strategies where students take more responsibility
for their own learning. One possible model is that of collaborative learning,
where students learn in groups through carefully crafted `problems', which has
long been suggested as a strategy for teaching statistics.
In this article, we review two specific approaches that fall under the
collaborative learning model: problem- and team-based learning. We consider the
evidence for changing to this model of teaching in statistics, as well as give
practical suggestions on how this could be implemented in typical statistics
classes in Higher Education.
| stat |
A Practical Response Adaptive Block Randomization Design with Analytic
Type I Error Protection | Response adaptive randomization is appealing in confirmatory adaptive
clinical trials from statistical, ethical, and pragmatic perspectives, in the
sense that subjects are more likely to be randomized to better performing
treatment groups based on accumulating data. The Doubly Adaptive Biased Coin
Design (DBCD) is a popular solution due to its asymptotic normal property of
final allocations, which further justifies its asymptotic type I error rate
control. As an alternative, we propose a Response Adaptive Block Randomization
(RABR) design with pre-specified randomization ratios for the control and
high-performing groups to robustly achieve desired final sample size per group
under different underlying responses, which is usually required in
industry-sponsored clinical studies. We show that the usual test statistic has
a controlled type I error rate. Our simulations further highlight the
advantages of the proposed design over the DBCD in terms of consistently
achieving final sample allocations and of power performance. We further apply
this design to a Phase III study evaluating the efficacy of two dosing regimens
of adjunctive everolimus in treating tuberous sclerosis complex but with no
previous dose-finding studies in this indication.
| stat |
Regression Analysis of Dependent Binary Data for Estimating Disease
Etiology from Case-Control Studies | In large-scale disease etiology studies, epidemiologists often need to use
multiple binary measures of unobserved causes of disease that are not perfectly
sensitive or specific to estimate cause-specific case fractions, referred to as
"population etiologic fractions" (PEFs). Despite recent methodological
advances, the scientific need of incorporating control data to estimate the
effect of explanatory variables upon the PEFs, however, remains unmet. In this
paper, we build on and extend nested partially-latent class model (npLCMs, Wu
et al., 2017) to a general framework for etiology regression analysis in
case-control studies. Data from controls provide requisite information about
measurement specificities and covariations, which is used to correctly assign
cause-specific probabilities for each case given her measurements. We estimate
the distribution of the controls' diagnostic measures given the covariates via
a separate regression model and a priori encourage simpler conditional
dependence structures. We use Markov chain Monte Carlo for posterior inference
of the PEF functions, cases' latent classes and the overall PEFs of policy
interest. We illustrate the regression analysis with simulations and show less
biased estimation and more valid inference of the overall PEFs than an npLCM
analysis omitting covariates. A regression analysis of data from a childhood
pneumonia study site reveals the dependence of pneumonia etiology upon season,
age, disease severity and HIV status.
| stat |
Bayesian Deep Convolutional Networks with Many Channels are Gaussian
Processes | There is a previously identified equivalence between wide fully connected
neural networks (FCNs) and Gaussian processes (GPs). This equivalence enables,
for instance, test set predictions that would have resulted from a fully
Bayesian, infinitely wide trained FCN to be computed without ever instantiating
the FCN, but by instead evaluating the corresponding GP. In this work, we
derive an analogous equivalence for multi-layer convolutional neural networks
(CNNs) both with and without pooling layers, and achieve state of the art
results on CIFAR10 for GPs without trainable kernels. We also introduce a Monte
Carlo method to estimate the GP corresponding to a given neural network
architecture, even in cases where the analytic form has too many terms to be
computationally feasible.
Surprisingly, in the absence of pooling layers, the GPs corresponding to CNNs
with and without weight sharing are identical. As a consequence, translation
equivariance, beneficial in finite channel CNNs trained with stochastic
gradient descent (SGD), is guaranteed to play no role in the Bayesian treatment
of the infinite channel limit - a qualitative difference between the two
regimes that is not present in the FCN case. We confirm experimentally, that
while in some scenarios the performance of SGD-trained finite CNNs approaches
that of the corresponding GPs as the channel count increases, with careful
tuning SGD-trained CNNs can significantly outperform their corresponding GPs,
suggesting advantages from SGD training compared to fully Bayesian parameter
estimation.
| stat |
Variational Noise-Contrastive Estimation | Unnormalised latent variable models are a broad and flexible class of
statistical models. However, learning their parameters from data is
intractable, and few estimation techniques are currently available for such
models. To increase the number of techniques in our arsenal, we propose
variational noise-contrastive estimation (VNCE), building on NCE which is a
method that only applies to unnormalised models. The core idea is to use a
variational lower bound to the NCE objective function, which can be optimised
in the same fashion as the evidence lower bound (ELBO) in standard variational
inference (VI). We prove that VNCE can be used for both parameter estimation of
unnormalised models and posterior inference of latent variables. The developed
theory shows that VNCE has the same level of generality as standard VI, meaning
that advances made there can be directly imported to the unnormalised setting.
We validate VNCE on toy models and apply it to a realistic problem of
estimating an undirected graphical model from incomplete data.
| stat |
Data-Driven Uncertainty Quantification and Propagation in Structural
Dynamics through a Hierarchical Bayesian Framework | In the presence of modeling errors, the mainstream Bayesian methods seldom
give a realistic account of uncertainties as they commonly underestimate the
inherent variability of parameters. This problem is not due to any
misconception in the Bayesian framework since it is absolutely robust with
respect to the modeling assumptions and the observed data. Rather, this issue
has deep roots in users' inability to develop an appropriate class of
probabilistic models. This paper bridges this significant gap, introducing a
novel Bayesian hierarchical setting, which breaks time-history vibrational
responses into several segments so as to capture and identify the variability
of inferred parameters over multiple segments. Since computation of the
posterior distributions in hierarchical models is expensive and cumbersome,
novel marginalization strategies, asymptotic approximations, and maximum a
posteriori estimations are proposed and outlined under a computational
algorithm aiming to handle both uncertainty quantification and propagation
tasks. For the first time, the connection between the ensemble covariance
matrix and hyper distribution parameters is characterized through approximate
estimations. Experimental and numerical examples are employed to illustrate the
efficacy and efficiency of the proposed method. It is observed that, when the
segments correspond to various system conditions and input characteristics, the
proposed method delivers robust parametric uncertainties with respect to
unknown phenomena such as ambient conditions, input characteristics, and
environmental factors.
| stat |
Semi-Supervised Non-Parametric Bayesian Modelling of Spatial Proteomics | Understanding sub-cellular protein localisation is an essential component to
analyse context specific protein function. Recent advances in quantitative
mass-spectrometry (MS) have led to high resolution mapping of thousands of
proteins to sub-cellular locations within the cell. Novel modelling
considerations to capture the complex nature of these data are thus necessary.
We approach analysis of spatial proteomics data in a non-parametric Bayesian
framework, using mixtures of Gaussian process regression models. The Gaussian
process regression model accounts for correlation structure within a
sub-cellular niche, with each mixture component capturing the distinct
correlation structure observed within each niche. Proteins with a priori
labelled locations motivate using semi-supervised learning to inform the
Gaussian process hyperparameters. We moreover provide an efficient
Hamiltonian-within-Gibbs sampler for our model. As in other recent work, we
reduce the computational burden associated with inversion of covariance
matrices by exploiting the structure in the covariance matrix. A tensor
decomposition of our covariance matrices allows extended Trench and Durbin
algorithms to be applied it order to reduce the computational complexity of
inversion and hence accelerate computation. A stand-alone R-package
implementing these methods using high-performance C++ libraries is available
at: https://github.com/ococrook/toeplitz
| stat |
Adaptive Approximate Bayesian Computation Tolerance Selection | Approximate Bayesian Computation (ABC) methods are increasingly used for
inference in situations in which the likelihood function is either
computationally costly or intractable to evaluate. Extensions of the basic ABC
rejection algorithm have improved the computational efficiency of the procedure
and broadened its applicability. The ABC-Population Monte Carlo (ABC-PMC)
approach of Beaumont et al. (2009) has become a popular choice for approximate
sampling from the posterior. ABC-PMC is a sequential sampler with an
iteratively decreasing value of the tolerance, which specifies how close the
simulated data need to be to the real data for acceptance. We propose a method
for adaptively selecting a sequence of tolerances that improves the
computational efficiency of the algorithm over other common techniques. In
addition we define a stopping rule as a by-product of the adaptation procedure,
which assists in automating termination of sampling. The proposed automatic
ABC-PMC algorithm can be easily implemented and we present several examples
demonstrating its benefits in terms of computational efficiency.
| stat |
Meaningful causal decompositions in health equity research: definition,
identification, and estimation through a weighting framework | Causal decomposition analyses can help build the evidence base for
interventions that address health disparities (inequities). They ask how
disparities in outcomes may change under hypothetical intervention. Through
study design and assumptions, they can rule out alternate explanations such as
confounding, selection-bias, and measurement error, thereby identifying
potential targets for intervention. Unfortunately, the literature on causal
decomposition analysis and related methods have largely ignored equity concerns
that actual interventionists would respect, limiting their relevance and
practical value. This paper addresses these concerns by explicitly considering
what covariates the outcome disparity and hypothetical intervention adjust for
(so-called allowable covariates) and the equity value judgements these choices
convey, drawing from the bioethics, biostatistics, epidemiology, and health
services research literatures. From this discussion, we generalize
decomposition estimands and formulae to incorporate allowable covariate sets,
to reflect equity choices, while still allowing for adjustment of non-allowable
covariates needed to satisfy causal assumptions. For these general formulae, we
provide weighting-based estimators based on adaptations of
ratio-of-mediator-probability and inverse-odds-ratio weighting. We discuss when
these estimators reduce to already used estimators under certain equity value
judgements, and a novel adaptation under other judgements.
| stat |
Quantile-Quantile Embedding for Distribution Transformation, Manifold
Embedding, and Image Embedding with Choice of Embedding Distribution | We propose a new embedding method, named Quantile-Quantile Embedding (QQE),
for distribution transformation, manifold embedding, and image embedding with
the ability to choose the embedding distribution. QQE, which uses the concept
of quantile-quantile plot from visual statistical tests, can transform the
distribution of data to any theoretical desired distribution or empirical
reference sample. Moreover, QQE gives the user a choice of embedding
distribution in embedding manifold of data into the low dimensional embedding
space. It can also be used for modifying the embedding distribution of
different dimensionality reduction methods, either basic or deep ones, for
better representation or visualization of data. We propose QQE in both
unsupervised and supervised manners. QQE can also transform the distribution to
either the exact reference distribution or shape of the reference distribution;
and one of its many applications is better discrimination of classes. Our
experiments on different synthetic and image datasets show the effectiveness of
the proposed embedding method.
| stat |
The numerical statistical fan for noisy experimental designs | Identifiability of polynomial models is a key requirement for multiple
regression. We consider an analogue of the so-called statistical fan, the set
of all maximal identifiable hierarchical models, for cases of noisy
experimental designs or measured covariate vectors with a given tolerance
vector. This gives rise to the definition of the numerical statistical fan. It
includes all maximal hierarchical models that avoid approximate linear
dependence of the model terms. We develop an algorithm to compute the numerical
statistical fan using recent results on the computation of all border bases of
a design ideal from the field of algebra. The ideas are applied to data from a
thermal spraying process. It turns out that the numerical statistical fan is
effectively computable and much smaller than the respective statistical fan.
The gained enhanced knowledge of the space of all stable identifiable
hierarchical models enables improved model selection procedures.
| stat |
Relative concentration bounds for the spectrum of kernel matrices | In this paper we study the concentration properties for the eigenvalues of
kernel matrices, which are central objects in a wide range of kernel methods
and, more recently, in network analysis. We present a set of concentration
inequalities tailored for each individual eigenvalue of the kernel matrix with
respect to its known asymptotic limit. The inequalities presented here are of
relative type, meaning that they scale with the eigenvalue in consideration,
which results in convergence rates that vary across the spectrum. The rates we
obtain here are faster than the typical $\O(\frac{1}{\sqrt n})$ and are often
exponential, depending on regularity assumptions of Sobolev type. One key
feature of our results is that they apply to non positive kernels, which is
fundamental in the context of network analysis. We show how our results are
well suited for the study of dot product kernels, which are related to random
geometric graphs on the sphere, via the graphon formalism. We illustrate our
results by applying them to a variety of dot product kernels on the sphere and
to the one dimensional Gaussian kernel.
| stat |
Efficient methods for the estimation of the multinomial parameter for
the two-trait group testing model | Estimation of a single Bernoulli parameter using pooled sampling is among the
oldest problems in the group testing literature. To carry out such estimation,
an array of efficient estimators have been introduced covering a wide range of
situations routinely encountered in applications. More recently, there has been
growing interest in using group testing to simultaneously estimate the joint
probabilities of two correlated traits using a multinomial model.
Unfortunately, basic estimation results, such as the maximum likelihood
estimator (MLE), have not been adequately addressed in the literature for such
cases. In this paper, we show that finding the MLE for this problem is
equivalent to maximizing a multinomial likelihood with a restricted parameter
space. A solution using the EM algorithm is presented which is guaranteed to
converge to the global maximizer, even on the boundary of the parameter space.
Two additional closed form estimators are presented with the goal of minimizing
the bias and/or mean square error. The methods are illustrated by considering
an application to the joint estimation of transmission prevalence for two
strains of the Potato virus Y by the aphid myzus persicae.
| stat |
The $\mathcal{F}$-family of covariance functions: A Mat\'ern analogue
for modeling random fields on spheres | The Mat{\'e}rn family of isotropic covariance functions has been central to
the theoretical development and application of statistical models for
geospatial data. For global data defined over the whole sphere representing
planet Earth, the natural distance between any two locations is the great
circle distance. In this setting, the Mat{\'e}rn family of covariance functions
has a restriction on the smoothness parameter, making it an unappealing choice
to model smooth data. Finding a suitable analogue for modelling data on the
sphere is still an open problem. This paper proposes a new family of isotropic
covariance functions for random fields defined over the sphere. The proposed
family has a parameter that indexes the mean square differentiability of the
corresponding Gaussian field, and allows for any admissible range of fractal
dimension. Our simulation study mimics the fixed domain asymptotic setting,
which is the most natural regime for sampling on a closed and bounded set. As
expected, our results support the analogous results (under the same asymptotic
scheme) for planar processes that not all parameters can be estimated
consistently. We apply the proposed model to a dataset of precipitable water
content over a large portion of the Earth, and show that the model gives more
precise predictions of the underlying process at unsampled locations than does
the Mat{\'e}rn model using chordal distances.
| stat |
A Mean-field Analysis of Deep ResNet and Beyond: Towards Provable
Optimization Via Overparameterization From Depth | Training deep neural networks with stochastic gradient descent (SGD) can
often achieve zero training loss on real-world tasks although the optimization
landscape is known to be highly non-convex. To understand the success of SGD
for training deep neural networks, this work presents a mean-field analysis of
deep residual networks, based on a line of works that interpret the continuum
limit of the deep residual network as an ordinary differential equation when
the network capacity tends to infinity. Specifically, we propose a new
continuum limit of deep residual networks, which enjoys a good landscape in the
sense that every local minimizer is global. This characterization enables us to
derive the first global convergence result for multilayer neural networks in
the mean-field regime. Furthermore, without assuming the convexity of the loss
landscape, our proof relies on a zero-loss assumption at the global minimizer
that can be achieved when the model shares a universal approximation property.
Key to our result is the observation that a deep residual network resembles a
shallow network ensemble, i.e. a two-layer network. We bound the difference
between the shallow network and our ResNet model via the adjoint sensitivity
method, which enables us to apply existing mean-field analyses of two-layer
networks to deep networks. Furthermore, we propose several novel training
schemes based on the new continuous model, including one training procedure
that switches the order of the residual blocks and results in strong empirical
performance on the benchmark datasets.
| stat |
portes: An R Package for Portmanteau Tests in Time Series Models | In this article, we introduce the R package portes with extensive
illustrative applications. The asymptotic distributions and the Monte Carlo
procedures of the most popular univariate and multivariate portmanteau test
statistics, including a new generalized variance statistic, for time series
models using the powerful parallel computing framework facility, are
implemented in this package. The proposed package has a general mechanism where
the user can compute the test statistic for the diagnostic checking of any time
series models. This package is also useful for simulating univariate and
multivariate seasonal and nonseasonal ARIMA/VARIMA time series with finite and
infinite variances, testing for stationarity and invertibility, and estimating
parameters from stable distributions.
| stat |
School value-added models for multivariate academic and non-academic
outcomes: A more rounded approach to using student data to inform school
accountability | Education systems around the world increasingly rely on school value-added
models to hold schools to account. These models typically focus on a limited
number of academic outcomes, failing to recognise the broader range of
non-academic student outcomes, attitudes and behaviours to which schools
contribute. We explore how the traditional multilevel modelling approach to
school value-added models can be extended to simultaneously analyse multiple
academic and non-academic outcomes and thereby can potentially provide a more
rounded approach to using student data to inform school accountability. We
jointly model student attainment, absence and exclusion data for schools in
England. We find different results across the three outcomes, in terms of the
size and consistency of school effects, and the importance of adjusting for
student and school characteristics. The results suggest the three outcomes are
capturing fundamentally distinct aspects of school performance, recommending
the consideration of non-academic outcomes in systems of school accountability.
| stat |
Conditional Modelling of Spatio-Temporal Extremes for Red Sea Surface
Temperatures | Recent extreme value theory literature has seen significant emphasis on the
modelling of spatial extremes, with comparatively little consideration of
spatio-temporal extensions. This neglects an important feature of extreme
events: their evolution over time. Many existing models for the spatial case
are limited by the number of locations they can handle; this impedes extension
to space-time settings, where models for higher dimensions are required.
Moreover, the spatio-temporal models that do exist are restrictive in terms of
the range of extremal dependence types they can capture. Recently, conditional
approaches for studying multivariate and spatial extremes have been proposed,
which enjoy benefits in terms of computational efficiency and an ability to
capture both asymptotic dependence and asymptotic independence. We extend this
class of models to a spatio-temporal setting, conditioning on the occurrence of
an extreme value at a single space-time location. We adopt a composite
likelihood approach for inference, which combines information from full
likelihoods across multiple space-time conditioning locations. We apply our
model to Red Sea surface temperatures, show that it fits well using a range of
diagnostic plots, and demonstrate how it can be used to assess the risk of
coral bleaching attributed to high water temperatures over consecutive days.
| stat |
Precise Unbiased Estimation in Randomized Experiments using Auxiliary
Observational Data | Randomized controlled trials (RCTs) are increasingly prevalent in education
research, and are often regarded as a gold standard of causal inference. Two
main virtues of randomized experiments are that they (1) do not suffer from
confounding, thereby allowing for an unbiased estimate of an intervention's
causal impact, and (2) allow for design-based inference, meaning that the
physical act of randomization largely justifies the statistical assumptions
made. However, RCT sample sizes are often small, leading to low precision; in
many cases RCT estimates may be too imprecise to guide policy or inform
science. Observational studies, by contrast, have strengths and weaknesses
complementary to those of RCTs. Observational studies typically offer much
larger sample sizes, but may suffer confounding. In many contexts, experimental
and observational data exist side by side, allowing the possibility of
integrating "big observational data" with "small but high-quality experimental
data" to get the best of both. Such approaches hold particular promise in the
field of education, where RCT sample sizes are often small due to cost
constraints, but automatic collection of observational data, such as in
computerized educational technology applications, or in state longitudinal data
systems (SLDS) with administrative data on hundreds of thousand of students,
has made rich, high-dimensional observational data widely available. We outline
an approach that allows one to employ machine learning algorithms to learn from
the observational data, and use the resulting models to improve precision in
randomized experiments. Importantly, there is no requirement that the machine
learning models are "correct" in any sense, and the final experimental results
are guaranteed to be exactly unbiased. Thus, there is no danger of confounding
biases in the observational data leaking into the experiment.
| stat |
Particle filters for high-dimensional geoscience applications: a review | Particle filters contain the promise of fully nonlinear data assimilation.
They have been applied in numerous science areas, but their application to the
geosciences has been limited due to their inefficiency in high-dimensional
systems in standard settings. However, huge progress has been made, and this
limitation is disappearing fast due to recent developments in proposal
densities, the use of ideas from (optimal) transportation, the use of
localisation and intelligent adaptive resampling strategies. Furthermore,
powerful hybrids between particle filters and ensemble Kalman filters and
variational methods have been developed. We present a state of the art
discussion of present efforts of developing particle filters for highly
nonlinear geoscience state-estimation problems with an emphasis on atmospheric
and oceanic applications, including many new ideas, derivations, and
unifications, highlighting hidden connections, and generating a valuable tool
and guide for the community. Initial experiments show that particle filters can
be competitive with present-day methods for numerical weather prediction
suggesting that they will become mainstream soon.
| stat |
Batch Self Organizing maps for distributional data using adaptive
distances | The paper deals with a Batch Self Organizing Map algorithm (DBSOM) for data
described by distributional-valued variables. This kind of variables is
characterized to take as values one-dimensional probability or frequency
distributions on a numeric support. The objective function optimized in the
algorithm depends on the choice of the distance measure. According to the
nature of the date, the $L_2$ Wasserstein distance is proposed as one of the
most suitable metrics to compare distributions. It is widely used in several
contexts of analysis of distributional data. Conventional batch SOM algorithms
consider that all variables are equally important for the training of the SOM.
However, it is well known that some variables are less relevant than others for
this task. In order to take into account the different contribution of the
variables we propose an adaptive version of the DBSOM algorithm that tackles
this problem with an additional step: a relevance weight is automatically
learned for each distributional-valued variable. Moreover, since the $L_2$
Wasserstein distance allows a decomposition into two components: one related to
the means and one related to the size and shape of the distributions, also
relevance weights are automatically learned for each of the measurement
components to emphasize the importance of the different estimated parameters of
the distributions. Examples of real and synthetic datasets of distributional
data illustrate the usefulness of the proposed DBSOM algorithms.
| stat |
Mixed Hamiltonian Monte Carlo for Mixed Discrete and Continuous
Variables | Hamiltonian Monte Carlo (HMC) has emerged as a powerful Markov Chain Monte
Carlo (MCMC) method to sample from complex continuous distributions. However, a
fundamental limitation of HMC is that it can not be applied to distributions
with mixed discrete and continuous variables. In this paper, we propose mixed
HMC (M-HMC) as a general framework to address this limitation. M-HMC is a novel
family of MCMC algorithms that evolves the discrete and continuous variables in
tandem, allowing more frequent updates of discrete variables while maintaining
HMC's ability to suppress random-walk behavior. We establish M-HMC's
theoretical properties, and present an efficient implementation with Laplace
momentum that introduces minimal overhead compared to existing HMC methods. The
superior performances of M-HMC over existing methods are demonstrated with
numerical experiments on Gaussian mixture models (GMMs), variable selection in
Bayesian logistic regression (BLR), and correlated topic models (CTMs).
| stat |
Wasserstein regularization for sparse multi-task regression | We focus in this paper on high-dimensional regression problems where each
regressor can be associated to a location in a physical space, or more
generally a generic geometric space. Such problems often employ sparse priors,
which promote models using a small subset of regressors. To increase
statistical power, the so-called multi-task techniques were proposed, which
consist in the simultaneous estimation of several related models. Combined with
sparsity assumptions, it lead to models enforcing the active regressors to be
shared across models, thanks to, for instance L1 / Lq norms. We argue in this
paper that these techniques fail to leverage the spatial information associated
to regressors. Indeed, while sparse priors enforce that only a small subset of
variables is used, the assumption that these regressors overlap across all
tasks is overly simplistic given the spatial variability observed in real data.
In this paper, we propose a convex regularizer for multi-task regression that
encodes a more flexible geometry. Our regularizer is based on unbalanced
optimal transport (OT) theory, and can take into account a prior geometric
knowledge on the regressor variables, without necessarily requiring overlapping
supports. We derive an efficient algorithm based on a regularized formulation
of OT, which iterates through applications of Sinkhorn's algorithm along with
coordinate descent iterations. The performance of our model is demonstrated on
regular grids with both synthetic and real datasets as well as complex
triangulated geometries of the cortex with an application in neuroimaging.
| stat |
How and Why Did Probability Theory Come About? | This paper is a top down historical perspective on the several phases in the
development of probability from its prehistoric origins to its modern day
evolution, as one of the key methodologies in artificial intelligence, data
science, and machine learning. It is written in honor of Barry Arnold's
birthday for his many contributions to statistical theory and methodology.
Despite the fact that much of Barry's work is technical, a descriptive document
to mark his achievements should not be viewed as being out of line. Barry's
dissertation adviser at Stanford (he received a Ph.D. in Statistics there) was
a philosopher of Science who dug deep in the foundations and roots of
probability, and it is this breadth of perspective is what Barry has inherent.
The paper is based on lecture materials compiled by the first author from
various published sources, and over a long period of time. The material below
gives a limited list of references, because the cast of characters is many, and
their contributions are a part of the historical heritage of those of us who
are interested in probability, statistics, and the many topics they have
spawned.
| stat |
Identifiability of Hierarchical Latent Attribute Models | Hierarchical Latent Attribute Models (HLAMs) are a family of discrete latent
variable models that are attracting increasing attention in educational,
psychological, and behavioral sciences. The key ingredients of an HLAM include
a binary structural matrix and a directed acyclic graph specifying hierarchical
constraints on the configurations of latent attributes. These components encode
practitioners' design information and carry important scientific meanings.
Despite the popularity of HLAMs, the fundamental identifiability issue remains
unaddressed. The existence of the attribute hierarchy graph leads to degenerate
parameter space, and the potentially unknown structural matrix further
complicates the identifiability problem. This paper addresses this issue of
identifying the latent structure and model parameters underlying an HLAM. We
develop sufficient and necessary identifiability conditions. These results
directly and sharply characterize the different impacts on identifiability cast
by different attribute types in the graph. The proposed conditions not only
provide insights into diagnostic test designs under the attribute hierarchy,
but also serve as tools to assess the validity of an estimated HLAM.
| stat |
Why optional stopping can be a problem for Bayesians | Recently, optional stopping has been a subject of debate in the Bayesian
psychology community. Rouder (2014) argues that optional stopping is no problem
for Bayesians, and even recommends the use of optional stopping in practice, as
do Wagenmakers et al. (2012). This article addresses the question whether
optional stopping is problematic for Bayesian methods, and specifies under
which circumstances and in which sense it is and is not. By slightly varying
and extending Rouder's (2014) experiments, we illustrate that, as soon as the
parameters of interest are equipped with default or pragmatic priors - which
means, in most practical applications of Bayes factor hypothesis testing -
resilience to optional stopping can break down. We distinguish between three
types of default priors, each having their own specific issues with optional
stopping, ranging from no-problem-at-all (Type 0 priors) to quite severe (Type
II priors).
| stat |
Robust Inference via Generative Classifiers for Handling Noisy Labels | Large-scale datasets may contain significant proportions of noisy (incorrect)
class labels, and it is well-known that modern deep neural networks (DNNs)
poorly generalize from such noisy training datasets. To mitigate the issue, we
propose a novel inference method, termed Robust Generative classifier (RoG),
applicable to any discriminative (e.g., softmax) neural classifier pre-trained
on noisy datasets. In particular, we induce a generative classifier on top of
hidden feature spaces of the pre-trained DNNs, for obtaining a more robust
decision boundary. By estimating the parameters of generative classifier using
the minimum covariance determinant estimator, we significantly improve the
classification accuracy with neither re-training of the deep model nor changing
its architectures. With the assumption of Gaussian distribution for features,
we prove that RoG generalizes better than baselines under noisy labels.
Finally, we propose the ensemble version of RoG to improve its performance by
investigating the layer-wise characteristics of DNNs. Our extensive
experimental results demonstrate the superiority of RoG given different
learning models optimized by several training techniques to handle diverse
scenarios of noisy labels.
| stat |
Estimating Unknown Cycles in Geophysical data | Examples of cyclic (periodic) behavior in geophysical data abound. In many
cases the primary period is known, such as in daily measurements of rain,
temperature, and sea level. However, many time series of measurements contain
cycles of unknown or varying length. We consider the problem of estimating the
unknown period in a time series.We review the basic methods, compare their
performance through a simulation studyusing observed sea level data, apply them
to an astronomical data set, and discuss generalizations of the methods.
| stat |
Double bootstrapping for visualising the distribution of descriptive
statistics of functional data | We propose a double bootstrap procedure for reducing coverage error in the
confidence intervals of descriptive statistics for independent and identically
distributed functional data. Through a series of Monte Carlo simulations, we
compare the finite sample performance of single and double bootstrap procedures
for estimating the distribution of descriptive statistics for independent and
identically distributed functional data. At the cost of longer computational
time, the double bootstrap with the same bootstrap method reduces confidence
level error and provides improved coverage accuracy than the single bootstrap.
Illustrated by a Canadian weather station data set, the double bootstrap
procedure presents a tool for visualising the distribution of the descriptive
statistics for the functional data.
| stat |
Infections Forecasting and Intervention Effect Evaluation for COVID-19
via a Data-Driven Markov Process and Heterogeneous Simulation | The Coronavirus Disease 2019 (COVID-19) pandemic has caused tremendous amount
of deaths and a devastating impact on the economic development all over the
world. Thus, it is paramount to control its further transmission, for which
purpose it is necessary to find the mechanism of its transmission process and
evaluate the effect of different control strategies. To deal with these issues,
we describe the transmission of COVID-19 as an explosive Markov process with
four parameters. The state transitions of the proposed Markov process can
clearly disclose the terrible explosion and complex heterogeneity of COVID-19.
Based on this, we further propose a simulation approach with heterogeneous
infections. Experimentations show that our approach can closely track the real
transmission process of COVID-19, disclose its transmission mechanism, and
forecast the transmission under different non-drug intervention strategies.
More importantly, our approach can helpfully develop effective strategies for
controlling COVID-19 and appropriately compare their control effect in
different countries/cities.
| stat |
Sparse multivariate regression with missing values and its application
to the prediction of material properties | In the field of materials science and engineering, statistical analysis and
machine learning techniques have recently been used to predict multiple
material properties from an experimental design. These material properties
correspond to response variables in the multivariate regression model. This
study conducts a penalized maximum likelihood procedure to estimate model
parameters, including the regression coefficients and covariance matrix of
response variables. In particular, we employ $l_1$-regularization to achieve a
sparse estimation of regression coefficients and the inverse covariance matrix
of response variables. In some cases, there may be a relatively large number of
missing values in response variables, owing to the difficulty in collecting
data on material properties. A method to improve prediction accuracy under the
situation with missing values incorporates a correlation structure among the
response variables into the statistical model. The expectation and maximization
algorithm is constructed, which enables application to a data set with missing
values in the responses. We apply our proposed procedure to real data
consisting of 22 material properties.
| stat |
Adaptive Sampling for Best Policy Identification in Markov Decision
Processes | We investigate the problem of best-policy identification in discounted Markov
Decision Processes (MDPs) when the learner has access to a generative model.
The objective is to devise a learning algorithm returning the best policy as
early as possible. We first derive a problem-specific lower bound of the sample
complexity satisfied by any learning algorithm. This lower bound corresponds to
an optimal sample allocation that solves a non-convex program, and hence, is
hard to exploit in the design of efficient algorithms. We then provide a simple
and tight upper bound of the sample complexity lower bound, whose corresponding
nearly-optimal sample allocation becomes explicit. The upper bound depends on
specific functionals of the MDP such as the sub-optimality gaps and the
variance of the next-state value function, and thus really captures the
hardness of the MDP. Finally, we devise KLB-TS (KL Ball Track-and-Stop), an
algorithm tracking this nearly-optimal allocation, and provide asymptotic
guarantees for its sample complexity (both almost surely and in expectation).
The advantages of KLB-TS against state-of-the-art algorithms are discussed and
illustrated numerically.
| stat |
Confidence interval for the AUC of SROC curve and some related methods
using bootstrap for meta-analysis of diagnostic accuracy studies | The area under the curve (AUC) of summary receiver operating characteristic
(SROC) curve is a primary statistical outcome for meta-analysis of diagnostic
test accuracy studies (DTA). However, its confidence interval has not been
reported in most of DTA meta-analyses, because no certain methods and
statistical packages have been provided. In this article, we provide a
bootstrap algorithm for computing the confidence interval of the AUC. Also,
using the bootstrap framework, we can conduct a bootstrap test for assessing
significance of the difference of AUCs for multiple diagnostic tests. In
addition, we provide an influence diagnostic method based on the AUC by
leave-one-study-out analyses. We present illustrative examples using two DTA
met-analyses for diagnostic tests of cervical cancer and asthma. We also
developed an easy-to-handle R package dmetatools for these computations. The
various quantitative evidence provided by these methods certainly supports the
interpretations and precise evaluations of statistical evidence of DTA
meta-analyses.
| stat |
Shrinkage Estimators in Online Experiments | We develop and analyze empirical Bayes Stein-type estimators for use in the
estimation of causal effects in large-scale online experiments. While online
experiments are generally thought to be distinguished by their large sample
size, we focus on the multiplicity of treatment groups. The typical analysis
practice is to use simple differences-in-means (perhaps with covariate
adjustment) as if all treatment arms were independent. In this work we develop
consistent, small bias, shrinkage estimators for this setting. In addition to
achieving lower mean squared error these estimators retain important
frequentist properties such as coverage under most reasonable scenarios. Modern
sequential methods of experimentation and optimization such as multi-armed
bandit optimization (where treatment allocations adapt over time to prior
responses) benefit from the use of our shrinkage estimators. Exploration under
empirical Bayes focuses more efficiently on near-optimal arms, improving the
resulting decisions made under uncertainty. We demonstrate these properties by
examining seventeen large-scale experiments conducted on Facebook from April to
June 2017.
| stat |
Modeling COVID-19 hospital admissions and occupancy in the Netherlands | We describe the models we built for hospital admissions and occupancy of
COVID-19 patients in the Netherlands. These models were used to make short-term
decisions about transfers of patients between regions and for long-term policy
making. We motivate and describe the model we used for predicting admissions
and how we use this to make predictions on occupancy.
| stat |
A race-DC in Big Data | The strategy of divide-and-combine (DC) has been widely used in the area of
big data. Bias-correction is crucial in the DC procedure for validly
aggregating the locally biased estimators, especial for the case when the
number of batches of data is large. This paper establishes a race-DC through a
residual-adjustment composition estimate (race). The race-DC applies to various
types of biased estimators, which include but are not limited to Lasso
estimator, Ridge estimator and principal component estimator in linear
regression, and least squares estimator in nonlinear regression. The resulting
global estimator is strictly unbiased under linear model, and is acceleratingly
bias-reduced in nonlinear model, and can achieve the theoretical optimality,
for the case when the number of batches of data is large. Moreover, the race-DC
is computationally simple because it is a least squares estimator in a pro
forma linear regression. Detailed simulation studies demonstrate that the
resulting global estimator is significantly bias-corrected, and the behavior is
comparable with the oracle estimation and is much better than the competitors.
| stat |
Learning a binary search with a recurrent neural network. A novel
approach to ordinal regression analysis | Deep neural networks are a family of computational models that are naturally
suited to the analysis of hierarchical data such as, for instance, sequential
data with the use of recurrent neural networks. In the other hand, ordinal
regression is a well-known predictive modelling problem used in fields as
diverse as psychometry to deep neural network based voice modelling. Their
specificity lies in the properties of their outcome variable, typically
considered as a categorical variable with natural ordering properties,
typically allowing comparisons between different states ("a little" is less
than "somewhat" which is itself less than "a lot", with transitivity allowed).
This article investigates the application of sequence-to-sequence learning
methods provided by the deep learning framework in ordinal regression, by
formulating the ordinal regression problem as a sequential binary search. A
method for visualizing the model's explanatory variables according to the
ordinal target variable is proposed, that bears some similarities to linear
discriminant analysis. The method is compared to traditional ordinal regression
methods on a number of benchmark dataset, and is shown to have comparable or
significantly better predictive power.
| stat |
Relaxing Bijectivity Constraints with Continuously Indexed Normalising
Flows | We show that normalising flows become pathological when used to model targets
whose supports have complicated topologies. In this scenario, we prove that a
flow must become arbitrarily numerically noninvertible in order to approximate
the target closely. This result has implications for all flow-based models, and
especially Residual Flows (ResFlows), which explicitly control the Lipschitz
constant of the bijection used. To address this, we propose Continuously
Indexed Flows (CIFs), which replace the single bijection used by normalising
flows with a continuously indexed family of bijections, and which can
intuitively "clean up" mass that would otherwise be misplaced by a single
bijection. We show theoretically that CIFs are not subject to the same
topological limitations as normalising flows, and obtain better empirical
performance on a variety of models and benchmarks.
| stat |
Nonparametric comparison of epidemic time trends: the case of COVID-19 | The COVID-19 pandemic is one of the most pressing issues at present. A
question which is particularly important for governments and policy makers is
the following: Does the virus spread in the same way in different countries? Or
are there significant differences in the development of the epidemic? In this
paper, we devise new inference methods that allow to detect differences in the
development of the COVID-19 epidemic across countries in a statistically
rigorous way. In our empirical study, we use the methods to compare the
outbreak patterns of the epidemic in a number of European countries.
| stat |
Comparing Probability Distributions with Conditional Transport | To measure the difference between two probability distributions, we propose
conditional transport (CT) as a new divergence and further approximate it with
the amortized CT (ACT) cost to make it amenable to implicit distributions and
stochastic gradient descent based optimization. ACT amortizes the computation
of its conditional transport plans and comes with unbiased sample gradients
that are straightforward to compute. When applied to train a generative model,
ACT is shown to strike a good balance between mode covering and seeking
behaviors and strongly resist mode collapse. On a wide variety of benchmark
datasets for generative modeling, substituting the default statistical distance
of an existing generative adversarial network with ACT is shown to consistently
improve the performance.
| stat |
Hierarchical Dynamic Modeling for Individualized Bayesian Forecasting | We present a case study and methodological developments in large-scale
hierarchical dynamic modeling for personalized prediction in commerce. The
context is supermarket sales, where improved forecasting of
customer/household-specific purchasing behavior informs decisions about
personalized pricing and promotions on a continuing basis. This is a big data,
big modeling and forecasting setting involving many thousands of customers and
items on sale, requiring sequential analysis, addressing information flows at
multiple levels over time, and with heterogeneity of customer profiles and item
categories. Models developed are fully Bayesian, interpretable and multi-scale,
with hierarchical forms overlaid on the inherent structure of the retail
setting. Customer behavior is modeled at several levels of aggregation, and
information flows from aggregate to individual levels. Forecasting at an
individual household level infers price sensitivity to inform personalized
pricing and promotion decisions. Methodological innovations include extensions
of Bayesian dynamic mixture models, their integration into multi-scale systems,
and forecast evaluation with context-specific metrics. The use of simultaneous
predictors from multiple hierarchical levels improves forecasts at the
customer-item level of main interest. This is evidenced across many different
households and items, indicating the utility of the modeling framework for this
and other individualized forecasting applications.
| stat |
Bayesian Nonparametric Nonhomogeneous Poisson Process with Applications
to USGS Earthquake Data | Intensity estimation is a common problem in statistical analysis of spatial
point pattern data. This paper proposes a nonparametric Bayesian method for
estimating the spatial point process intensity based on mixture of finite
mixture (MFM) model. MFM approach leads to a consistent estimate of the
intensity of spatial point patterns in different areas while considering
heterogeneity. An efficient Markov chain Monte Carlo (MCMC) algorithm is
proposed for our method. Extensive simulation studies are carried out to
examine empirical performance of the proposed method. The usage of our proposed
method is further illustrated with the analysis of the Earthquake Hazards
Program of United States Geological Survey (USGS) earthquake data.
| stat |
Continuum centroid classifier for functional data | Aiming at the binary classification of functional data, we propose the
continuum centroid classifier (CCC) built upon projections of functional data
onto one specific direction. This direction is obtained via bridging the
regression and classification. Controlling the extent of supervision, our
technique is neither unsupervised nor fully supervised. Thanks to the intrinsic
infinite dimension of functional data, one of two subtypes of CCC enjoys the
(asymptotic) zero misclassification rate. Our proposal includes an effective
algorithm that yields a consistent empirical counterpart of CCC. Simulation
studies demonstrate the performance of CCC in different scenarios. Finally, we
apply CCC to two real examples.
| stat |
A semiparametric instrumental variable approach to optimal treatment
regimes under endogeneity | There is a fast-growing literature on estimating optimal treatment regimes
based on randomized trials or observational studies under a key identifying
condition of no unmeasured confounding. Because confounding by unmeasured
factors cannot generally be ruled out with certainty in observational studies
or randomized trials subject to noncompliance, we propose a general
instrumental variable approach to learning optimal treatment regimes under
endogeneity. Specifically, we establish identification of both value function
$E[Y_{\mathcal{D}(L)}]$ for a given regime $\mathcal{D}$ and optimal regimes
$\text{argmax}_{\mathcal{D}} E[Y_{\mathcal{D}(L)}]$ with the aid of a binary
instrumental variable, when no unmeasured confounding fails to hold. We also
construct novel multiply robust classification-based estimators. Furthermore,
we propose to identify and estimate optimal treatment regimes among those who
would comply to the assigned treatment under a standard monotonicity
assumption. In this latter case, we establish the somewhat surprising result
that complier optimal regimes can be consistently estimated without directly
collecting compliance information and therefore without the complier average
treatment effect itself being identified. Our approach is illustrated via
extensive simulation studies and a data application on the effect of child
rearing on labor participation.
| stat |
Robust Sample Weighting to Facilitate Individualized Treatment Rule
Learning for a Target Population | Learning individualized treatment rules (ITRs) is an important topic in
precision medicine. Current literature mainly focuses on deriving ITRs from a
single source population. We consider the observational data setting when the
source population differs from a target population of interest. We assume
subject covariates are available from both populations, but treatment and
outcome data are only available from the source population. Although adjusting
for differences between source and target populations can potentially lead to
an improved ITR for the target population, it can substantially increase the
variability in ITR estimation. To address this dilemma, we develop a weighting
framework that aims to tailor an ITR for a given target population and protect
against high variability due to superfluous covariate shift adjustments. Our
method seeks covariate balance over a nonparametric function class
characterized by a reproducing kernel Hilbert space and can improve many ITR
learning methods that rely on weights. We show that the proposed method
encompasses importance weights and the so-called overlap weights as two extreme
cases, allowing for a better bias-variance trade-off in between. Numerical
examples demonstrate that the use of our weighting method can greatly improve
ITR estimation for the target population compared with other weighting methods.
| stat |
Distributions.jl: Definition and Modeling of Probability Distributions
in the JuliaStats Ecosystem | Random variables and their distributions are a central part in many areas of
statistical methods. The Distributions.jl package provides Julia users and
developers tools for working with probability distributions, leveraging Julia
features for their intuitive and flexible manipulation, while remaining highly
efficient through zero-cost abstractions.
| stat |
Robust estimation via generalized quasi-gradients | We explore why many recently proposed robust estimation problems are
efficiently solvable, even though the underlying optimization problems are
non-convex. We study the loss landscape of these robust estimation problems,
and identify the existence of "generalized quasi-gradients". Whenever these
quasi-gradients exist, a large family of low-regret algorithms are guaranteed
to approximate the global minimum; this includes the commonly-used filtering
algorithm.
For robust mean estimation of distributions under bounded covariance, we show
that any first-order stationary point of the associated optimization problem is
an {approximate global minimum} if and only if the corruption level $\epsilon <
1/3$. Consequently, any optimization algorithm that aproaches a stationary
point yields an efficient robust estimator with breakdown point $1/3$. With
careful initialization and step size, we improve this to $1/2$, which is
optimal.
For other tasks, including linear regression and joint mean and covariance
estimation, the loss landscape is more rugged: there are stationary points
arbitrarily far from the global minimum. Nevertheless, we show that generalized
quasi-gradients exist and construct efficient algorithms. These algorithms are
simpler than previous ones in the literature, and for linear regression we
improve the estimation error from $O(\sqrt{\epsilon})$ to the optimal rate of
$O(\epsilon)$ for small $\epsilon$ assuming certified hypercontractivity. For
mean estimation with near-identity covariance, we show that a simple gradient
descent algorithm achieves breakdown point $1/3$ and iteration complexity
$\tilde{O}(d/\epsilon^2)$.
| stat |
A statistical theory of cold posteriors in deep neural networks | To get Bayesian neural networks to perform comparably to standard neural
networks it is usually necessary to artificially reduce uncertainty using a
"tempered" or "cold" posterior. This is extremely concerning: if the prior is
accurate, Bayes inference/decision theory is optimal, and any artificial
changes to the posterior should harm performance. While this suggests that the
prior may be at fault, here we argue that in fact, BNNs for image
classification use the wrong likelihood. In particular, standard image
benchmark datasets such as CIFAR-10 are carefully curated. We develop a
generative model describing curation which gives a principled Bayesian account
of cold posteriors, because the likelihood under this new generative model
closely matches the tempered likelihoods used in past work.
| stat |
One-class classification with application to forensic analysis | The analysis of broken glass is forensically important to reconstruct the
events of a criminal act. In particular, the comparison between the glass
fragments found on a suspect (recovered cases) and those collected on the crime
scene (control cases) may help the police to correctly identify the
offender(s). The forensic issue can be framed as a one-class classification
problem. One-class classification is a recently emerging and special
classification task, where only one class is fully known (the so-called target
class), while information on the others is completely missing. We propose to
consider classic Gini's transvariation probability as a measure of typicality,
i.e. a measure of resemblance between an observation and a set of well-known
objects (the control cases). The aim of the proposed Transvariation-based
One-Class Classifier (TOCC) is to identify the best boundary around the target
class, that is, to recognise as many target objects as possible while rejecting
all those deviating from this class.
| stat |
Surrogate-assisted reliability-based design optimization: a survey and a
new general framework | Reliability-based design optimization (RBDO) is an active field of research
with an ever increasing number of contributions. Numerous methods have been
proposed for the solution of RBDO, a complex problem that combines optimization
and reliability analysis. Classical approaches are based on approximation
methods and have been classified in review papers. In this paper, we first
review classical approaches based on approximation methods such as FORM, and
also more recent methods that rely upon surrogate modelling and Monte Carlo
simulation. We then propose a general framework for the solution of RBDO
problems that includes three independent blocks, namely adaptive surrogate
modelling, reliability analysis and optimization. These blocks are
non-intrusive with respect to each other and can be plugged independently in
the framework. After a discussion on numerical considerations that require
attention for the framework to yield robust solutions to various types of
problems, the latter is applied to three examples (using two analytical
functions and a finite element model). Kriging and support vector machines
together with their own active learning schemes are considered in the surrogate
model block. In terms of reliability analysis, the proposed framework is
illustrated using both crude Monte Carlo and subset simulation. Finally, the
covariance-matrix adaptation - evolution scheme (CMA-ES), a global search
algorithm, or sequential quadratic programming (SQP), a local gradient-based
method, are used in the optimization block. The comparison of the results to
benchmark studies show the effectiveness and efficiency of the proposed
framework.
| stat |
Estimating Extreme Value Index by Subsampling for Massive Datasets with
Heavy-Tailed Distributions | Modern statistical analyses often encounter datasets with massive sizes and
heavy-tailed distributions. For datasets with massive sizes, traditional
estimation methods can hardly be used to estimate the extreme value index
directly. To address the issue, we propose here a subsampling-based method.
Specifically, multiple subsamples are drawn from the whole dataset by using the
technique of simple random subsampling with replacement. Based on each
subsample, an approximate maximum likelihood estimator can be computed. The
resulting estimators are then averaged to form a more accurate one. Under
appropriate regularity conditions, we show theoretically that the proposed
estimator is consistent and asymptotically normal. With the help of the
estimated extreme value index, a normal range can be established for a
heavy-tailed random variable. Observations that fall outside the range should
be treated as suspected records and can be practically regarded as outliers.
Extensive simulation experiments are provided to demonstrate the promising
performance of our method. A real data analysis is also presented for
illustration purpose.
| stat |
Accelerating proximal Markov chain Monte Carlo by using an explicit
stabilised method | We present a highly efficient proximal Markov chain Monte Carlo methodology
to perform Bayesian computation in imaging problems. Similarly to previous
proximal Monte Carlo approaches, the proposed method is derived from an
approximation of the Langevin diffusion. However, instead of the conventional
Euler-Maruyama approximation that underpins existing proximal Monte Carlo
methods, here we use a state-of-the-art orthogonal Runge-Kutta-Chebyshev
stochastic approximation that combines several gradient evaluations to
significantly accelerate its convergence speed, similarly to accelerated
gradient optimisation methods. The proposed methodology is demonstrated via a
range of numerical experiments, including non-blind image deconvolution,
hyperspectral unmixing, and tomographic reconstruction, with total-variation
and $\ell_1$-type priors. Comparisons with Euler-type proximal Monte Carlo
methods confirm that the Markov chains generated with our method exhibit
significantly faster convergence speeds, achieve larger effective sample sizes,
and produce lower mean square estimation errors at equal computational budget.
| stat |
A Conway-Maxwell-Poisson GARMA Model for Count Data | We propose a flexible model for count time series which has potential uses
for both underdispersed and overdispersed data. The model is based on the
Conway-Maxwell-Poisson (COM-Poisson) distribution with parameters varying along
time to take serial correlation into account. Model estimation is challenging
however and require the application of recently proposed methods to deal with
the intractable normalising constant as well as efficiently sampling values
from the COM-Poisson distribution.
| stat |
A Fully Bayesian Approach to Assessment of Model Adequacy in Inverse
Problems | We consider the problem of assessing goodness of fit of a single Bayesian
model to the observed data in the inverse problem context. A novel procedure of
goodness of fit test is proposed, based on construction of reference
distributions using the `inverse' part of the given model. This is motivated by
an example from palaeoclimatology in which it is of interest to reconstruct
past climates using information obtained from fossils deposited in lake
sediment.
Technically, given a model $f(Y\mid X,\theta)$, where $Y$ is the observed
data and $X$ is a set of (non-random) covariates, we obtain reference
distributions based on the posterior $\pi(\tilde X\mid Y)$, where $\tilde X$
must be interpreted as the {\it unobserved} random vector corresponding to the
{\it observed} covariates $X$. Put simply, if the posterior distribution
$\pi(\tilde X\mid Y)$ gives high density to the observed covariates $X$, or
equivalently, if the posterior distribution of $T(\tilde X)$ gives high density
to $T(X)$, where $T$ is any appropriate statistic, then we say that the model
fits the data. Otherwise the model in question is not adequate. We provide
decision-theoretic justification of our proposed approach and discuss other
theoretical and computational advantages. We demonstrate our methodology with
many simulated examples and three complex, high-dimensional, realistic
palaeoclimate problems, including the motivating palaeoclimate problem.
| stat |
Relaxed Wasserstein with Applications to GANs | Wasserstein Generative Adversarial Networks (WGANs) provide a versatile class
of models, which have attracted great attention in various applications.
However, this framework has two main drawbacks: (i) Wasserstein-1 (or
Earth-Mover) distance is restrictive such that WGANs cannot always fit data
geometry well; (ii) It is difficult to achieve fast training of WGANs. In this
paper, we propose a new class of \textit{Relaxed Wasserstein} (RW) distances by
generalizing Wasserstein-1 distance with Bregman cost functions. We show that
RW distances achieve nice statistical properties while not sacrificing the
computational tractability. Combined with the GANs framework, we develop
Relaxed WGANs (RWGANs) which are not only statistically flexible but can be
approximated efficiently using heuristic approaches. Experiments on real images
demonstrate that the RWGAN with Kullback-Leibler (KL) cost function outperforms
other competing approaches, e.g., WGANs, even with gradient penalty.
| stat |
Beyond Strictly Proper Scoring Rules: The Importance of Being Local | The evaluation of probabilistic forecasts plays a central role both in the
interpretation and in the use of forecast systems and their development.
Probabilistic scores (scoring rules) provide statistical measures to assess the
quality of probabilistic forecasts. Often, many probabilistic forecast systems
are available while evaluations of their performance are not standardized, with
different scoring rules being used to measure different aspects of forecast
performance. Even when the discussion is restricted to strictly proper scoring
rules, there remains considerable variability between them; indeed strictly
proper scoring rules need not rank competing forecast systems in the same order
when none of these systems are perfect. The locality property is explored to
further distinguish scoring rules. The nonlocal strictly proper scoring rules
considered are shown to have a property that can produce "unfortunate"
evaluations. Particularly the fact that Continuous Rank Probability Score
prefers the outcome close to the median of the forecast distribution regardless
the probability mass assigned to the value at/near the median raises concern to
its use. The only local strictly proper scoring rules, the logarithmic score,
has direct interpretations in terms of probabilities and bits of information.
The nonlocal strictly proper scoring rules, on the other hand, lack meaningful
direct interpretation for decision support. The logarithmic score is also shown
to be invariant under smooth transformation of the forecast variable, while the
nonlocal strictly proper scoring rules considered may, however, change their
preferences due to the transformation. It is therefore suggested that the
logarithmic score always be included in the evaluation of probabilistic
forecasts.
| stat |
Fast Parameter Inference in a Biomechanical Model of the Left Ventricle
using Statistical Emulation | A central problem in biomechanical studies of personalised human left
ventricular (LV) modelling is estimating the material properties and
biophysical parameters from in-vivo clinical measurements in a time frame
suitable for use within a clinic. Understanding these properties can provide
insight into heart function or dysfunction and help inform personalised
medicine. However, finding a solution to the differential equations which
mathematically describe the kinematics and dynamics of the myocardium through
numerical integration can be computationally expensive. To circumvent this
issue, we use the concept of emulation to infer the myocardial properties of a
healthy volunteer in a viable clinical time frame using in-vivo magnetic
resonance image (MRI) data. Emulation methods avoid computationally expensive
simulations from the LV model by replacing the biomechanical model, which is
defined in terms of explicit partial differential equations, with a surrogate
model inferred from simulations generated before the arrival of a patient,
vastly improving computational efficiency at the clinic. We compare and
contrast two emulation strategies: (i) emulation of the computational model
outputs and (ii) emulation of the loss between the observed patient data and
the computational model outputs. These strategies are tested with two different
interpolation methods, as well as two different loss functions...
| stat |
Functional continuum regression | Functional principal component regression (PCR) can fail to provide good
prediction if the response is highly correlated with some excluded functional
principal component(s). This situation is common since the construction of
functional principal components never involves the response. Aiming at this
shortcoming, we develop functional continuum regression (CR). The framework of
functional CR includes, as special cases, both functional PCR and functional
partial least squares (PLS). Functional CR is expected to own a better accuracy
than functional PCR and functional PLS both in estimation and prediction;
evidence for this is provided through simulation and numerical case studies.
Also, we demonstrate the consistency of estimators given by functional CR.
| stat |
Testing the first-order separability hypothesis for spatio-temporal
point patterns | First-order separability of a spatio-temporal point process plays a
fundamental role in the analysis of spatio-temporal point pattern data. While
it is often a convenient assumption that simplifies the analysis greatly,
existing non-separable structures should be accounted for in the model
construction. We propose three different tests to investigate this hypothesis
as a step of preliminary data analysis. The first two tests are exact or
asymptotically exact for Poisson processes. The first test based on
permutations and global envelopes allows us to detect at which spatial and
temporal locations or lags the data deviate from the null hypothesis. The
second test is a simple and computationally cheap $\chi^2$-test. The third test
is based on statistical reconstruction method and can be generally applied for
non-Poisson processes. The performance of the first two tests is studied in a
simulation study for Poisson and non-Poisson models. The third test is applied
to the real data of the UK 2001 epidemic foot and mouth disease.
| stat |
Boltzmann Exploration Expectation-Maximisation | We present a general method for fitting finite mixture models (FMM). Learning
in a mixture model consists of finding the most likely cluster assignment for
each data-point, as well as finding the parameters of the clusters themselves.
In many mixture models, this is difficult with current learning methods, where
the most common approach is to employ monotone learning algorithms e.g. the
conventional expectation-maximisation algorithm. While effective, the success
of any monotone algorithm is crucially dependant on good parameter
initialisation, where a common choice is $K$-means initialisation, commonly
employed for Gaussian mixture models.
For other types of mixture models, the path to good initialisation parameters
is often unclear and may require a problem-specific solution. To this end, we
propose a general heuristic learning algorithm that utilises Boltzmann
exploration to assign each observation to a specific base distribution within
the mixture model, which we call Boltzmann exploration expectation-maximisation
(BEEM). With BEEM, hard assignments allow straight forward parameter learning
for each base distribution by conditioning only on its assigned observations.
Consequently, it can be applied to mixtures of any base distribution where
single component parameter learning is tractable. The stochastic learning
procedure is able to escape local optima and is thus insensitive to parameter
initialisation. We show competitive performance on a number of synthetic
benchmark cases as well as on real-world datasets.
| stat |
Individual-centered partial information in social networks | Most existing statistical network analysis literature assumes a global view
of the network, under which community detection, testing, and other statistical
procedures are developed. Yet in the real world, people frequently make
decisions based on their partial understanding of the network information. As
individuals barely know beyond friends' friends, we assume that an individual
of interest knows all paths of length up to $L=2$ that originate from her. As a
result, this individual's perceived adjacency matrix $\bbB$ differs
significantly from the usual adjacency matrix $\bbA$ based on the global
information. The new individual-centered partial information framework sparks
an array of interesting endeavors from theory to practice. Key general
properties on the eigenvalues and eigenvectors of $\bbB_E$, a major term of
$\bbB$, are derived. These general results, coupled with the classic stochastic
block model, lead to a new theory-backed spectral approach to detecting the
community memberships based on an anchored individual's partial information.
Real data analysis delivers interesting insights that cannot be obtained from
global network analysis.
| stat |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.