title
stringlengths 6
244
| abstract
stringlengths 19
5.09k
| label
stringclasses 10
values |
|---|---|---|
Debiased Bayesian inference for average treatment effects | Bayesian approaches have become increasingly popular in causal inference
problems due to their conceptual simplicity, excellent performance and in-built
uncertainty quantification ('posterior credible sets'). We investigate Bayesian
inference for average treatment effects from observational data, which is a
challenging problem due to the missing counterfactuals and selection bias.
Working in the standard potential outcomes framework, we propose a data-driven
modification to an arbitrary (nonparametric) prior based on the propensity
score that corrects for the first-order posterior bias, thereby improving
performance. We illustrate our method for Gaussian process (GP) priors using
(semi-)synthetic data. Our experiments demonstrate significant improvement in
both estimation accuracy and uncertainty quantification compared to the
unmodified GP, rendering our approach highly competitive with the
state-of-the-art.
| stat |
Nonlinear Time Series Classification Using Bispectrum-based Deep
Convolutional Neural Networks | Time series classification using novel techniques has experienced a recent
resurgence and growing interest from statisticians, subject-domain scientists,
and decision makers in business and industry. This is primarily due to the ever
increasing amount of big and complex data produced as a result of technological
advances. A motivating example is that of Google trends data, which exhibit
highly nonlinear behavior. Although a rich literature exists for addressing
this problem, existing approaches mostly rely on first and second order
properties of the time series, since they typically assume linearity of the
underlying process. Often, these are inadequate for effective classification of
nonlinear time series data such as Google Trends data. Given these
methodological deficiencies and the abundance of nonlinear time series that
persist among real-world phenomena, we introduce an approach that merges higher
order spectral analysis (HOSA) with deep convolutional neural networks (CNNs)
for classifying time series. The effectiveness of our approach is illustrated
using simulated data and two motivating industry examples that involve Google
trends data and electronic device energy consumption data.
| stat |
Forecasting Intra-Hour Imbalances in Electric Power Systems | Keeping the electricity production in balance with the actual demand is
becoming a difficult and expensive task in spite of an involvement of
experienced human operators. This is due to the increasing complexity of the
electric power grid system with the intermittent renewable production as one of
the contributors. A beforehand information about an occurring imbalance can
help the transmission system operator to adjust the production plans, and thus
ensure a high security of supply by reducing the use of costly balancing
reserves, and consequently reduce undesirable fluctuations of the 50 Hz power
system frequency. In this paper, we introduce the relatively new problem of an
intra-hour imbalance forecasting for the transmission system operator (TSO). We
focus on the use case of the Norwegian TSO, Statnett. We present a
complementary imbalance forecasting tool that is able to support the TSO in
determining the trend of future imbalances, and show the potential to
proactively alleviate imbalances with a higher accuracy compared to the
contemporary solution.
| stat |
Understanding partition comparison indices based on counting object
pairs | In unsupervised machine learning, agreement between partitions is commonly
assessed with so-called external validity indices. Researchers tend to use and
report indices that quantify agreement between two partitions for all clusters
simultaneously. Commonly used examples are the Rand index and the adjusted Rand
index. Since these overall measures give a general notion of what is going on,
their values are usually hard to interpret.
Three families of indices based on counting object pairs are analyzed. It is
shown that the overall indices can be decomposed into indices that reflect the
degree of agreement on the level of individual clusters. The overall indices
based on the pair-counting approach are sensitive to cluster size imbalance:
they tend to reflect the degree of agreement on the large clusters and provide
little to no information on smaller clusters. Furthermore, the value of
Rand-like indices is determined to a large extent by the number of pairs of
objects that are not joined in either of the partitions.
| stat |
Tree-Wasserstein Barycenter for Large-Scale Multilevel Clustering and
Scalable Bayes | We study in this paper a variant of Wasserstein barycenter problem, which we
refer to as tree-Wasserstein barycenter, by leveraging a specific class of
ground metrics, namely tree metrics, for Wasserstein distance. Drawing on the
tree structure, we propose an efficient algorithmic approach to solve the
tree-Wasserstein barycenter and its variants. The proposed approach is not only
fast for computation but also efficient for memory usage. Exploiting the
tree-Wasserstein barycenter and its variants, we scale up multi-level
clustering and scalable Bayes, especially for large-scale applications where
the number of supports in probability measures is large. Empirically, we test
our proposed approach against other baselines on large-scale synthetic and real
datasets.
| stat |
Jump balls, rating falls, and elite status: A sensitivity analysis of
three quarterback rating statistics | Quarterback performance can be difficult to rank, and much effort has been
spent in creating new rating systems. However, the input statistics for such
ratings are subject to randomness and factors outside the quarterback's
control. To investigate this variance, we perform a sensitivity analysis of
three quarterback rating statistics: the Traditional 1971 rating by Smith, the
Burke, and the Wages of Wins ratings. The comparisons are made at the team
level for the 32 NFL teams from 2002-2015, thus giving each case an even 16
games. We compute quarterback ratings for each offense with 1-5 additional
touchdowns, 1-5 fewer interceptions, 1-5 additional sacks, and a 1-5 percent
increase in the passing completion rate. Our sensitivity analysis provides
insight into whether an elite passing team could seem mediocre or vice versa
based on random outcomes. The results indicate that the Traditional rating is
the most sensitive statistic with respect to touchdowns, interceptions, and
completions, whereas the Burke rating is most sensitive to sacks. The analysis
suggests that team passing offense rankings are highly sensitive to aspects of
football that are out of the quarterback's hands (e.g., deflected passes that
lead to interceptions). Thus, on the margins, we show arguments about whether a
specific quarterback has entered the elite or remains mediocre are irrelevant.
| stat |
Rediscovering a little known fact about the t-test: algebraic,
geometric, distributional and graphical considerations | We discuss the role that the null hypothesis should play in the construction
of a test statistic used to make a decision about that hypothesis. To construct
the test statistic for a point null hypothesis about a binomial proportion, a
common recommendation is to act as if the null hypothesis is true. We argue
that, on the surface, the one-sample t-test of a point null hypothesis about a
Gaussian population mean does not appear to follow the recommendation. We show
how simple algebraic manipulations of the usual t-statistic lead to an
equivalent test procedure consistent with the recommendation, we provide
geometric intuition regarding this equivalence, and we consider extensions to
testing nested hypotheses in Gaussian linear models. We discuss an application
to graphical residual diagnostics where the form of the test statistic makes a
practical difference. We argue that these issues should be discussed in
advanced undergraduate and graduate courses.
| stat |
A Simple Algorithm for Exact Multinomial Tests | This work proposes a new method for computing acceptance regions of exact
multinomial tests. From this an algorithm is derived, which finds exact
p-values for tests of simple multinomial hypotheses. Using concepts from
discrete convex analysis, the method is proven to be exact for various popular
test statistics, including Pearson's chi-square and the log-likelihood ratio.
The proposed algorithm improves greatly on the naive approach using full
enumeration of the sample space. However, its use is limited to multinomial
distributions with a small number of categories, as the runtime grows
exponentially in the number of possible outcomes.
The method is applied in a simulation study and uses of multinomial tests in
forecast evaluation are outlined. Additionally, properties of a test statistic
using probability ordering, referred to as the "exact multinomial test" by some
authors, are investigated and discussed. The algorithm is implemented in the
accompanying R package ExactMultinom.
| stat |
Asymptotically Exact and Fast Gaussian Copula Models for Imputation of
Mixed Data Types | Missing values with mixed data types is a common problem in a large number of
machine learning applications such as processing of surveys and in different
medical applications. Recently, Gaussian copula models have been suggested as a
means of performing imputation of missing values using a probabilistic
framework. While the present Gaussian copula models have shown to yield state
of the art performance, they have two limitations: they are based on an
approximation that is fast but may be imprecise and they do not support
unordered multinomial variables. We address the first limitation using direct
and arbitrarily precise approximations both for model estimation and imputation
by using randomized quasi-Monte Carlo procedures. The method we provide has
lower errors for the estimated model parameters and the imputed values,
compared to previously proposed methods. We also extend the previous Gaussian
copula models to include unordered multinomial variables in addition to the
present support of ordinal, binary, and continuous variables.
| stat |
Estimation of large block structured covariance matrices: Application to
"multi-omic" approaches to study seed quality | Motivated by an application in high-throughput genomics and metabolomics, we
propose a novel, efficient and fully data-driven approach for estimating large
block structured sparse covariance matrices in the case where the number of
variables is much larger than the number of samples without limiting ourselves
to block diagonal matrices. Our approach consists in approximating such a
covariance matrix by the sum of a low-rank sparse matrix and a diagonal matrix.
Our methodology also can deal with matrices for which the block structure
appears only if the columns and rows are permuted according to an unknown
permutation. Our technique is implemented in the R package \texttt{BlockCov}
which is available from the Comprehensive R Archive Network (CRAN) and from
GitHub. In order to illustrate the statistical and numerical performance of our
package some numerical experiments are provided as well as a thorough
comparison with alternative methods. Finally, our approach is applied to the
use of "multi-omic" approaches for studying seed quality.
| stat |
A Descriptive Study of Variable Discretization and Cost-Sensitive
Logistic Regression on Imbalanced Credit Data | Training classification models on imbalanced data tends to result in bias
towards the majority class. In this paper, we demonstrate how variable
discretization and cost-sensitive logistic regression help mitigate this bias
on an imbalanced credit scoring dataset, and further show the application of
the variable discretization technique on the data from other domains,
demonstrating its potential as a generic technique for classifying imbalanced
data beyond credit socring. The performance measurements include ROC curves,
Area under ROC Curve (AUC), Type I Error, Type II Error, accuracy, and F1
score. The results show that proper variable discretization and cost-sensitive
logistic regression with the best class weights can reduce the model bias
and/or variance. From the perspective of the algorithm, cost-sensitive logistic
regression is beneficial for increasing the value of predictors even if they
are not in their optimized forms while maintaining monotonicity. From the
perspective of predictors, the variable discretization performs better than
cost-sensitive logistic regression, provides more reasonable coefficient
estimates for predictors which have nonlinear relationships against their
empirical logit, and is robust to penalty weights on misclassifications of
events and non-events determined by their apriori proportions.
| stat |
Two-directional simultaneous inference for high-dimensional models | This paper proposes a general two directional simultaneous inference (TOSI)
framework for high-dimensional models with a manifest variable or latent
variable structure, for example, high-dimensional mean models, high-dimensional
sparse regression models, and high-dimensional latent factors models. TOSI
performs simultaneous inference on a set of parameters from two directions, one
to test whether the assumed zero parameters indeed are zeros and one to test
whether exist zeros in the parameter set of nonzeros. As a result, we can
exactly identify whether the parameters are zeros, thereby keeping the data
structure fully and parsimoniously expressed. We theoretically prove that the
proposed TOSI method asymptotically controls the Type I error at the
prespecified significance level and that the testing power converges to one.
Simulations are conducted to examine the performance of the proposed method in
finite sample situations and two real datasets are analyzed. The results show
that the TOSI method is more predictive and has more interpretable estimators
than existing methods.
| stat |
What Does the "Mean" Really Mean? | The arithmetic average of a collection of observed values of a homogeneous
collection of quantities is often taken to be the most representative
observation. There are several arguments supporting this choice the moment of
inertia being the most familiar. But what does this mean?
In this note, we bring forth the Kolmogorov-Nagumo point of view that the
arithmetic average is a special case of a sequence of functions of a special
kind, the quadratic and the geometric means being some of the other cases. The
median fails to belong to this class of functions. The Kolmogorov-Nagumo
interpretation is the most defensible and the most definitive one for the
arithmetic average, but its essence boils down to the fact that this average is
merely an abstraction which has meaning only within its mathematical set-up.
| stat |
Analysis of air pollution time series using complexity-invariant
distance and information measures | Air pollution is known to be a major threat for human and ecosystem health. A
proper understanding of the factors generating pollution and of the behavior of
air pollution in time is crucial to support the development of effective
policies aiming at the reduction of pollutant concentration. This paper
considers the hourly time series of three pollutants, namely NO$_2$, O$_3$ and
PM$_{2.5}$, collected on sixteen measurement stations in Switzerland. The air
pollution patterns due to the location of measurement stations and their
relationship with anthropogenic activities, and specifically land use, are
studied using two approaches: Fisher-Shannon information plane and
complexity-invariant distance between time series. A clustering analysis is
used to recognize within the measurements of a same pollutant group of stations
behaving in a similar way. The results clearly demonstrate the relationship
between the air pollution probability densities and land use activities.
| stat |
Exploiting locality in high-dimensional factorial hidden Markov models | We propose algorithms for approximate filtering and smoothing in
high-dimensional factorial hidden Markov models. The approximation involves
discarding, in a principled way, likelihood factors according a notion of
locality in a factor graph associated with the emission distribution. This
allows the exponential-in-dimension cost of exact filtering and smoothing to be
avoided. We prove that the approximation accuracy, measured in a local total
variation norm, is `dimension-free' in the sense that as the overall dimension
of the model increases the error bounds we derive do not necessarily degrade. A
key step in the analysis is to quantify the error introduced by localizing the
likelihood function in a Bayes' rule update. The factorial structure of the
likelihood function which we exploit arises naturally when data have known
spatial or network structure. We demonstrate the new algorithms on synthetic
examples and a London Underground passenger flow problem, where the factor
graph is effectively given by the train network.
| stat |
Correcting Predictions for Approximate Bayesian Inference | Bayesian models quantify uncertainty and facilitate optimal decision-making
in downstream applications. For most models, however, practitioners are forced
to use approximate inference techniques that lead to sub-optimal decisions due
to incorrect posterior predictive distributions. We present a novel approach
that corrects for inaccuracies in posterior inference by altering the
decision-making process. We train a separate model to make optimal decisions
under the approximate posterior, combining interpretable Bayesian modeling with
optimization of direct predictive accuracy in a principled fashion. The
solution is generally applicable as a plug-in module for predictive
decision-making for arbitrary probabilistic programs, irrespective of the
posterior inference strategy. We demonstrate the approach empirically in
several problems, confirming its potential.
| stat |
Influences in Forecast Errors for Wind and Photovoltaic Power: A Study
on Machine Learning Models | Despite the increasing importance of forecasts of renewable energy, current
planning studies only address a general estimate of the forecast quality to be
expected and selected forecast horizons.
However, these estimates allow only a limited and highly uncertain use in the
planning of electric power distribution. More reliable planning processes
require considerably more information about future forecast quality.
In this article, we present an in-depth analysis and comparison of
influencing factors regarding uncertainty in wind and photovoltaic power
forecasts, based on four different machine learning (ML) models.
In our analysis, we found substantial differences in uncertainty depending on
ML models, data coverage, and seasonal patterns that have to be considered in
future planning studies.
| stat |
Nishimori meets Bethe: a spectral method for node classification in
sparse weighted graphs | This article unveils a new relation between the Nishimori temperature
parametrizing a distribution P and the Bethe free energy on random Erdos-Renyi
graphs with edge weights distributed according to P. Estimating the Nishimori
temperature being a task of major importance in Bayesian inference problems, as
a practical corollary of this new relation, a numerical method is proposed to
accurately estimate the Nishimori temperature from the eigenvalues of the Bethe
Hessian matrix of the weighted graph. The algorithm, in turn, is used to
propose a new spectral method for node classification in weighted (possibly
sparse) graphs. The superiority of the method over competing state-of-the-art
approaches is demonstrated both through theoretical arguments and real-world
data experiments.
| stat |
Good distribution modelling with the R package good | Although models for count data with over-dispersion have been widely
considered in the literature, models for under-dispersion -- the opposite
phenomenon -- have received less attention as it is only relatively common in
particular research fields such as biodosimetry and ecology. The Good
distribution is a flexible alternative for modelling count data showing either
over-dispersion or under-dispersion, although no R packages are still available
to the best of our knowledge. We aim to present in the following the R package
good that computes the standard probabilistic functions (i.e., probability
density function, cumulative distribution function, and quantile function) and
generates random samples from a population following a Good distribution. The
package also considers a function for Good regression, including covariates in
a similar way to that of the standard glm function. We finally show the use of
such a package with some real-world data examples addressing both
over-dispersion and especially under-dispersion.
| stat |
PLSO: A generative framework for decomposing nonstationary timeseries
into piecewise stationary oscillatory components | To capture the slowly time-varying spectral content of real-world
time-series, a common paradigm is to partition the data into approximately
stationary intervals and perform inference in the time-frequency domain.
However, this approach lacks a corresponding nonstationary time-domain
generative model for the entire data and thus, time-domain inference occurs in
each interval separately. This results in distortion/discontinuity around
interval boundaries and, consequently, can lead to erroneous inferences based
on any quantities derived from the posterior, such as the phase. To address
these shortcomings, we propose the Piecewise Locally Stationary Oscillation
(PLSO) generative model for decomposing time-series data with slowly
time-varying spectra into several oscillatory, piecewise-stationary processes.
PLSO, being a nonstationary time-domain generative model, enables inference on
the entire time-series, without boundary effects, and, at the same time,
provides a characterization of its time-varying spectral properties. For
inference, we propose a novel two-stage algorithm that combines Kalman theory
and an accelerated proximal gradient algorithm for the nonconvex objective. We
demonstrate these points through experiments on simulated data and real neural
data from the rat and the human brain.
| stat |
Training-image based geostatistical inversion using a spatial generative
adversarial neural network | Probabilistic inversion within a multiple-point statistics framework is often
computationally prohibitive for high-dimensional problems. To partly address
this, we introduce and evaluate a new training-image based inversion approach
for complex geologic media. Our approach relies on a deep neural network of the
generative adversarial network (GAN) type. After training using a training
image (TI), our proposed spatial GAN (SGAN) can quickly generate 2D and 3D
unconditional realizations. A key characteristic of our SGAN is that it defines
a (very) low-dimensional parameterization, thereby allowing for efficient
probabilistic inversion using state-of-the-art Markov chain Monte Carlo (MCMC)
methods. In addition, available direct conditioning data can be incorporated
within the inversion. Several 2D and 3D categorical TIs are first used to
analyze the performance of our SGAN for unconditional geostatistical
simulation. Training our deep network can take several hours. After training,
realizations containing a few millions of pixels/voxels can be produced in a
matter of seconds. This makes it especially useful for simulating many
thousands of realizations (e.g., for MCMC inversion) as the relative cost of
the training per realization diminishes with the considered number of
realizations. Synthetic inversion case studies involving 2D steady-state flow
and 3D transient hydraulic tomography with and without direct conditioning data
are used to illustrate the effectiveness of our proposed SGAN-based inversion.
For the 2D case, the inversion rapidly explores the posterior model
distribution. For the 3D case, the inversion recovers model realizations that
fit the data close to the target level and visually resemble the true model
well.
| stat |
Statistical significance in high-dimensional linear mixed models | This paper concerns the development of an inferential framework for
high-dimensional linear mixed effect models. These are suitable models, for
instance, when we have $n$ repeated measurements for $M$ subjects. We consider
a scenario where the number of fixed effects $p$ is large (and may be larger
than $M$), but the number of random effects $q$ is small. Our framework is
inspired by a recent line of work that proposes de-biasing penalized estimators
to perform inference for high-dimensional linear models with fixed effects
only. In particular, we demonstrate how to correct a `naive' ridge estimator in
extension of work by B\"uhlmann (2013) to build asymptotically valid confidence
intervals for mixed effect models. We validate our theoretical results with
numerical experiments, in which we show our method outperforms those that fail
to account for correlation induced by the random effects. For a practical
demonstration we consider a riboflavin production dataset that exhibits group
structure, and show that conclusions drawn using our method are consistent with
those obtained on a similar dataset without group structure.
| stat |
A Bayesian nonparametric approach to the approximation of the global
stable manifold | We propose a Bayesian nonparametric model based on Markov Chain Monte Carlo
(MCMC) methods for unveiling the structure of the invariant global stable
manifold from observed time-series data. The underlying unknown dynamical
process is possibly contaminated by additive noise. We introduce the Stable
Manifold Geometric Stick Breaking Reconstruction (SM-GSBR) model with which we
reconstruct the unknown dynamic equations and in parallel we estimate the
global structure of the perturbed stable manifold. Our method works for
noninvertible maps without modifications. The stable manifold estimation
procedure is demonstrated specifically in the case of polynomial maps.
Simulations based on synthetic time series are presented.
| stat |
Efficient Estimation in Single Index Models through Smoothing splines | We consider estimation and inference in a single index regression model with
an unknown but smooth link function. In contrast to the standard approach of
using kernels or regression splines, we use smoothing splines to estimate the
smooth link function. We develop a method to compute the penalized least
squares estimators (PLSEs) of the parametric and the nonparametric components
given independent and identically distributed (i.i.d.)~data. We prove the
consistency and find the rates of convergence of the estimators. We establish
asymptotic normality under under mild assumption and prove asymptotic
efficiency of the parametric component under homoscedastic errors. A finite
sample simulation corroborates our asymptotic theory. We also analyze a car
mileage data set and a Ozone concentration data set. The identifiability and
existence of the PLSEs are also investigated.
| stat |
Discovering a Regularity: the Case of An 800-year Law of Advances in
Small-Arms Technologies | Considering a broad family of technologies where a measure of performance
(MoP) is difficult or impossible to formulate, we seek an alternative measure
that exhibits a regular pattern of evolution over time, similar to how a MoP
may follow a Moore's law. In an empirical case study, we explore an approach to
identifying such a composite measure called a Figure of Regularity (FoR). We
use the proposed approach to identify a novel FoR for diverse classes of small
arms - bows, crossbows, harquebuses, muskets, rifles, repeaters, and assault
rifles - and show that this FoR agrees well with the empirical data. We
identify a previously unreported regular trend in the FoR of an exceptionally
long duration - from approximately 1200 CE to the present - and discuss how
research managers can analyze long-term trends in conjunction with a portfolio
of research directions.
| stat |
Design-Based Inference for Spatial Experiments with Interference | We consider design-based causal inference in settings where randomized
treatments have effects that bleed out into space in complex ways that overlap
and in violation of the standard "no interference" assumption for many causal
inference methods. We define a spatial "average marginalized response," which
characterizes how, in expectation, units of observation that are a specified
distance from an intervention point are affected by treatments at that point,
averaging over effects emanating from other intervention points. We establish
conditions for non-parametric identification, asymptotic distributions of
estimators, and recovery of structural effects. We propose methods for both
sample-theoretic and permutation-based inference. We provide illustrations
using randomized field experiments on forest conservation and health.
| stat |
Testing for a Change in Mean After Changepoint Detection | While many methods are available to detect structural changes in a time
series, few procedures are available to quantify the uncertainty of these
estimates post-detection. In this work, we fill this gap by proposing a new
framework to test the null hypothesis that there is no change in mean around an
estimated changepoint. We further show that it is possible to efficiently carry
out this framework in the case of changepoints estimated by binary segmentation
and its variants, $\ell_{0}$ segmentation, or the fused lasso. Our setup allows
us to condition on much less information than existing approaches, which yields
higher powered tests. We apply our proposals in a simulation study and on a
dataset of chromosomal guanine-cytosine content. These approaches are freely
available in the R package ChangepointInference at
https://jewellsean.github.io/changepoint-inference/.
| stat |
Binary Classification of Gaussian Mixtures: Abundance of Support
Vectors, Benign Overfitting and Regularization | Deep neural networks generalize well despite being exceedingly
overparameterized and being trained without explicit regularization. This
curious phenomenon, often termed benign overfitting, has inspired extensive
research activity in establishing its statistical principles: Under what
conditions is the phenomenon observed? How do these depend on the data and on
the training algorithm? When does regularization benefit generalization? While
these questions remain wide open for deep neural nets, recent works have
attempted gaining insights by studying simpler, often linear, models. Our paper
contributes to this growing line of work by examining binary linear
classification under the popular generative Gaussian mixture model. Motivated
by recent results on the implicit bias of gradient descent, we study both
max-margin SVM classifiers (corresponding to logistic loss) and min-norm
interpolating classifiers (corresponding to least-squares loss). First, we
leverage an idea introduced in [V. Muthukumar et al., arXiv:2005.08054, (2020)]
to relate the SVM solution to the least-squares (LS) interpolating solution.
Second, we derive novel non-asymptotic bounds on the classification error of
the LS solution. Combining the two, we present novel sufficient conditions on
the overparameterization ratio and on the signal-to-noise ratio (SNR) for
benign overfitting to occur. Contrary to previously studied discriminative data
models, our results emphasize the crucial role of the SNR. Moreover, we
investigate the role of regularization and identify precise conditions under
which the interpolating estimator performs better than the regularized
estimates. We corroborate our theoretical findings with numerical simulations.
| stat |
Hamiltonian Monte Carlo with Energy Conserving Subsampling | Hamiltonian Monte Carlo (HMC) samples efficiently from high-dimensional
posterior distributions with proposed parameter draws obtained by iterating on
a discretized version of the Hamiltonian dynamics. The iterations make HMC
computationally costly, especially in problems with large datasets, since it is
necessary to compute posterior densities and their derivatives with respect to
the parameters. Naively computing the Hamiltonian dynamics on a subset of the
data causes HMC to lose its key ability to generate distant parameter proposals
with high acceptance probability. The key insight in our article is that
efficient subsampling HMC for the parameters is possible if both the dynamics
and the acceptance probability are computed from the same data subsample in
each complete HMC iteration. We show that this is possible to do in a
principled way in a HMC-within-Gibbs framework where the subsample is updated
using a pseudo marginal MH step and the parameters are then updated using an
HMC step, based on the current subsample. We show that our subsampling methods
are fast and compare favorably to two popular sampling algorithms that utilize
gradient estimates from data subsampling. We also explore the current
limitations of subsampling HMC algorithms by varying the quality of the
variance reducing control variates used in the estimators of the posterior
density and its gradients.
| stat |
Phase transition in a power-law uniform hypergraph | We propose a power-law $m$-uniform random hypergraph on $n$ vertexes. In this
hypergraph, each vertex is independently assigned a random weight from a
power-law distribution with exponent $\alpha\in(0,\infty)$ and the hyperedge
probabilities are defined as functions of the random weights. We characterize
the number of hyperedge and the number of loose 2-cycle. There is a phase
transition phenomenon for the number of hyperedge at $\alpha=1$. Interestingly,
for the number of loose 2-cycle, phase transition occurs at both $\alpha=1$ and
$\alpha=2$.
| stat |
On Weighted Multivariate Sign Functions | Multivariate sign functions are often used for robust estimation and
inference. We propose using data dependent weights in association with such
functions. The proposed weighted sign functions retain desirable robustness
properties, while significantly improving efficiency in estimation and
inference compared to unweighted multivariate sign-based methods. Using
weighted signs, we demonstrate methods of robust location estimation and robust
principal component analysis. We extend the scope of using robust multivariate
methods to include robust sufficient dimension reduction and functional outlier
detection. Several numerical studies and real data applications demonstrate the
efficacy of the proposed methodology.
| stat |
A Semiparametric Approach to Model-based Sensitivity Analysis in
Observational Studies | When drawing causal inference from observational data, there is always
concern about unmeasured confounding. One way to tackle this is to conduct a
sensitivity analysis. One widely-used sensitivity analysis framework
hypothesizes the existence of a scalar unmeasured confounder U and asks how the
causal conclusion would change were U measured and included in the primary
analysis. Works along this line often make various parametric assumptions on U,
for the sake of mathematical and computational simplicity. In this article, we
substantively further this line of research by developing a valid sensitivity
analysis that leaves the distribution of U unrestricted. Our semiparametric
estimator has three desirable features compared to many existing methods in the
literature. First, our method allows for a larger and more flexible family of
models, and mitigates observable implications (Franks et al., 2019). Second,
our methods work seamlessly with any primary analysis that models the outcome
regression parametrically. Third, our method is easy to use and interpret. We
construct both pointwise confidence intervals and confidence bands that are
uniformly valid over a given sensitivity parameter space, thus formally
accounting for unknown sensitivity parameters. We apply our proposed method on
an influential yet controversial study of the causal relationship between war
experiences and political activeness using observational data from Uganda.
| stat |
k-hop Graph Neural Networks | Graph neural networks (GNNs) have emerged recently as a powerful architecture
for learning node and graph representations. Standard GNNs have the same
expressive power as the Weisfeiler-Leman test of graph isomorphism in terms of
distinguishing non-isomorphic graphs. However, it was recently shown that this
test cannot identify fundamental graph properties such as connectivity and
triangle freeness. We show that GNNs also suffer from the same limitation. To
address this limitation, we propose a more expressive architecture, k-hop GNNs,
which updates a node's representation by aggregating information not only from
its direct neighbors, but from its k-hop neighborhood. We show that the
proposed architecture can identify fundamental graph properties. We evaluate
the proposed architecture on standard node classification and graph
classification datasets. Our experimental evaluation confirms our theoretical
findings since the proposed model achieves performance better or comparable to
standard GNNs and to state-of-the-art algorithms.
| stat |
Estimation of treatment effects following a sequential trial of multiple
treatments | When a clinical trial is subject to a series of interim analyses as a result
of which the study may be terminated or modified, final frequentist analyses
need to take account of the design used. Failure to do so may result in
overstated levels of significance, biased effect estimates and confidence
intervals with inadequate coverage probabilities. A wide variety of valid
methods of frequentist analysis have been devised for sequential designs
comparing a single experimental treatment with a single control treatment. It
is less clear how to perform the final analysis of a sequential or adaptive
design applied in a more complex setting, for example to determine which
treatment or set of treatments amongst several candidates should be
recommended.
This paper has been motivated by consideration of a trial in which four
treatments for sepsis are to be compared, with interim analyses allowing the
dropping of treatments or termination of the trial to declare a single winner
or to conclude that there is little difference between the treatments that
remain. The approach taken is based on the method of Rao-Blackwellisation which
enhances the accuracy of unbiased estimates available from the first interim
analysis by taking their conditional expectations given final sufficient
statistics. Analytic approaches to determine such expectations are difficult
and specific to the details of the design, and instead "reverse simulations"
are conducted to construct replicate realisations of the first interim analysis
from the final test statistics. The method also provides approximate confidence
intervals for the differences between treatments.
| stat |
Parameter inference for a stochastic kinetic model of expanded
polyglutamine proteins | The presence of protein aggregates in cells is a known feature of many human
age-related diseases, such as Huntington's disease. Simulations using fixed
parameter values in a model of the dynamic evolution of expanded polyglutamine
(PolyQ) proteins in cells have been used to gain a better understanding of the
biological system, how to focus drug development and how to construct more
efficient designs of future laboratory-based in vitro experiments. However,
there is considerable uncertainty about the values of some of the parameters
governing the system. Currently, appropriate values are chosen by ad hoc
attempts to tune the parameters so that the model output matches experimental
data. The problem is further complicated by the fact that the data only offer a
partial insight into the underlying biological process: the data consist only
of the proportions of cell death and of cells with inclusion bodies at a few
time points, corrupted by measurement error.
Developing inference procedures to estimate the model parameters in this
scenario is a significant task. The model probabilities corresponding to the
observed proportions cannot be evaluated exactly and so they are estimated
within the inference algorithm by repeatedly simulating realisations from the
model. In general such an approach is computationally very expensive and we
therefore construct Gaussian process emulators for the key quantities and
reformulate our algorithm around these fast stochastic approximations. We
conclude by examining the fit of our model and highlight appropriate values of
the model parameters leading to new insights into the underlying biological
processes such as the kinetics of aggregation.
| stat |
Soft Maximin Estimation for Heterogeneous Array Data | The extraction of a common signal across many recordings is difficult when
each recording -- in addition to the signal -- contains large, unique variation
components. Maximin estimation has previously been proposed as a robust
estimation method in the presence of heterogeneous noise.
We propose soft maximin estimation as a computationally attractive
methodology for estimating a common signal from heterogeneous data. The soft
maximin loss is introduced as an aggregation, controlled by a parameter
$\zeta>0$, of explained variances and the estimator is obtained by minimizing
the penalized soft maximin loss.
By establishing statistical and computational properties we argue that the
soft maximin method is a statistically sensibel and computationally attractive
alternative to existing methods. In particular we demonstrate, on simulated and
real data, that the soft maximin estimator can outperform existing methods both
in terms of predictive performance and run time. We also provide a time and
memory efficient implementation for data with array-tensor structure in the R
package SMMA available on CRAN.
| stat |
Iterative Weak Learnability and Multi-Class AdaBoost | We construct an efficient recursive ensemble algorithm for the multi-class
classification problem, inspired by SAMME (Zhu, Zou, Rosset, and Hastie
(2009)). We strengthen the weak learnability condition in Zhu, Zou, Rosset, and
Hastie (2009) by requiring that the weak learnability condition holds for any
subset of labels with at least two elements. This condition is simpler to check
than many proposed alternatives (e.g., Mukherjee and Schapire (2013)). As
SAMME, our algorithm is reduced to the Adaptive Boosting algorithm (Schapire
and Freund (2012)) if the number of labels is two, and can be motivated as a
functional version of the steepest descending method to find an optimal
solution. In contrast to SAMME, our algorithm's final hypothesis converges to
the correct label with probability 1. For any number of labels, the probability
of misclassification vanishes exponentially as the training period increases.
The sum of the training error and an additional term, that depends only on the
sample size, bounds the generalization error of our algorithm as the Adaptive
Boosting algorithm.
| stat |
Power law dynamics in genealogical graphs | Several populational networks present complex topologies when implemented in
evolutionary algorithms. A common feature of these topologies is the emergence
of a power law. In genealogical networks, the power law can be observed by
measuring the impact of individuals in the population, which can be calculated
through the Event Takeover Value (ETV) algorithm. In this paper, we show
evidence that the different power-law deviations, resulting from the ETV
distributions of genealogical graphs, are static images of a dynamic evolution
that can be well described by $q$-exponential distribution.
| stat |
Universal Hypothesis Testing with Kernels: Asymptotically Optimal Tests
for Goodness of Fit | We characterize the asymptotic performance of nonparametric goodness of fit
testing. The exponential decay rate of the type-II error probability is used as
the asymptotic performance metric, and a test is optimal if it achieves the
maximum rate subject to a constant level constraint on the type-I error
probability. We show that two classes of Maximum Mean Discrepancy (MMD) based
tests attain this optimality on $\mathbb R^d$, while the quadratic-time Kernel
Stein Discrepancy (KSD) based tests achieve the maximum exponential decay rate
under a relaxed level constraint. Under the same performance metric, we proceed
to show that the quadratic-time MMD based two-sample tests are also optimal for
general two-sample problems, provided that kernels are bounded continuous and
characteristic. Key to our approach are Sanov's theorem from large deviation
theory and the weak metrizable properties of the MMD and KSD.
| stat |
Elastic analysis of irregularly or sparsely sampled curves | We provide statistical analysis methods for samples of curves when the image
but not the parametrisation of the curves is of interest. A parametrisation
invariant analysis can be based on the elastic distance of the curves modulo
warping, but existing methods have limitations in common realistic settings
where curves are irregularly and potentially sparsely observed. We provide
methods and algorithms to approximate the elastic distance for such curves via
interpreting them as polygons. Moreover, we propose to use spline curves for
modelling smooth or polygonal Fr\'echet means of open or closed curves with
respect to the elastic distance and show identifiability of the spline model
modulo warping. We illustrate the use of our methods for elastic mean and
distance computation by application to two datasets. The first application
clusters sparsely sampled GPS tracks based on the elastic distance and computes
smooth means for each cluster to find new paths on Tempelhof field in Berlin.
The second classifies irregularly sampled handwritten spirals of Parkinson's
patients and controls based on the elastic distance to a mean spiral curve
computed using our approach. All developed methods are implemented in the
\texttt{R}-package \texttt{elasdics} and evaluated in simulations.
| stat |
Density estimation and modeling on symmetric spaces | In many applications, data and/or parameters are supported on non-Euclidean
manifolds. It is important to take into account the geometric structure of
manifolds in statistical analysis to avoid misleading results. Although there
has been a considerable focus on simple and specific manifolds, there is a lack
of general and easy-to-implement statistical methods for density estimation and
modeling on manifolds. In this article, we consider a very broad class of
manifolds: non-compact Riemannian symmetric spaces. For this class, we provide
a very general mathematical result for easily calculating volume changes of the
exponential and logarithm map between the tangent space and the manifold. This
allows one to define statistical models on the tangent space, push these models
forward onto the manifold, and easily calculate induced distributions by
Jacobians. To illustrate the statistical utility of this theoretical result, we
provide a general method to construct distributions on symmetric spaces. In
particular, we define the log-Gaussian distribution as an analogue of the
multivariate Gaussian distribution in Euclidean space. With these new kernels
on symmetric spaces, we also consider the problem of density estimation. Our
proposed approach can use any existing density estimation approach designed for
Euclidean spaces and push it forward to the manifold with an easy-to-calculate
adjustment. We provide theorems showing that the induced density estimators on
the manifold inherit the statistical optimality properties of the parent
Euclidean density estimator; this holds for both frequentist and Bayesian
nonparametric methods. We illustrate the theory and practical utility of the
proposed approach on the space of positive definite matrices.
| stat |
A Bootstrap Method for Error Estimation in Randomized Matrix
Multiplication | In recent years, randomized methods for numerical linear algebra have
received growing interest as a general approach to large-scale problems.
Typically, the essential ingredient of these methods is some form of randomized
dimension reduction, which accelerates computations, but also creates random
approximation error. In this way, the dimension reduction step encodes a
tradeoff between cost and accuracy. However, the exact numerical relationship
between cost and accuracy is typically unknown, and consequently, it may be
difficult for the user to precisely know (1) how accurate a given solution is,
or (2) how much computation is needed to achieve a given level of accuracy. In
the current paper, we study randomized matrix multiplication (sketching) as a
prototype setting for addressing these general problems. As a solution, we
develop a bootstrap method for \emph{directly estimating} the accuracy as a
function of the reduced dimension (as opposed to deriving worst-case bounds on
the accuracy in terms of the reduced dimension). From a computational
standpoint, the proposed method does not substantially increase the cost of
standard sketching methods, and this is made possible by an "extrapolation"
technique. In addition, we provide both theoretical and empirical results to
demonstrate the effectiveness of the proposed method.
| stat |
Dropout as a Structured Shrinkage Prior | Dropout regularization of deep neural networks has been a mysterious yet
effective tool to prevent overfitting. Explanations for its success range from
the prevention of "co-adapted" weights to it being a form of cheap Bayesian
inference. We propose a novel framework for understanding multiplicative noise
in neural networks, considering continuous distributions as well as Bernoulli
noise (i.e. dropout). We show that multiplicative noise induces structured
shrinkage priors on a network's weights. We derive the equivalence through
reparametrization properties of scale mixtures and without invoking any
approximations. Given the equivalence, we then show that dropout's Monte Carlo
training objective approximates marginal MAP estimation. We leverage these
insights to propose a novel shrinkage framework for resnets, terming the prior
'automatic depth determination' as it is the natural analog of automatic
relevance determination for network depth. Lastly, we investigate two inference
strategies that improve upon the aforementioned MAP approximation in regression
benchmarks.
| stat |
A User-Friendly Computational Framework for Robust Structured Regression
Using the L$_2$ Criterion | We introduce a user-friendly computational framework for implementing robust
versions of a wide variety of structured regression methods using the L$_{2}$
criterion. In addition to introducing a scalable algorithm for performing
L$_{2}$E regression, our framework also enables robust regression using the
L$_{2}$ criterion for additional structural constraints, works without
requiring complex tuning procedures, can be used to automatically identify
heterogeneous subpopulations, and can incorporate readily available non-robust
structured regression solvers. We provide convergence guarantees for the
framework and demonstrate its flexibility with some examples.
| stat |
Assessing Algorithmic Fairness with Unobserved Protected Class Using
Data Combination | The increasing impact of algorithmic decisions on people's lives compels us
to scrutinize their fairness and, in particular, the disparate impacts that
ostensibly-color-blind algorithms can have on different groups. Examples
include credit decisioning, hiring, advertising, criminal justice, personalized
medicine, and targeted policymaking, where in some cases legislative or
regulatory frameworks for fairness exist and define specific protected classes.
In this paper we study a fundamental challenge to assessing disparate impacts
in practice: protected class membership is often not observed in the data. This
is particularly a problem in lending and healthcare. We consider the use of an
auxiliary dataset, such as the US census, to construct models that predict the
protected class from proxy variables, such as surname and geolocation. We show
that even with such data, a variety of common disparity measures are generally
unidentifiable, providing a new perspective on the documented biases of popular
proxy-based methods. We provide exact characterizations of the
tightest-possible set of all possible true disparities that are consistent with
the data (and possibly any assumptions). We further provide optimization-based
algorithms for computing and visualizing these sets and statistical tools to
assess sampling uncertainty. Together, these enable reliable and robust
assessments of disparities -- an important tool when disparity assessment can
have far-reaching policy implications. We demonstrate this in two case studies
with real data: mortgage lending and personalized medicine dosing.
| stat |
Forecasting confirmed cases of the COVID-19 pandemic with a
migration-based epidemiological model | The unprecedented coronavirus disease 2019 (COVID-19) pandemic is still a
worldwide threat to human life since its invasion into the daily lives of the
public in the first several months of 2020. Predicting the size of confirmed
cases is important for countries and communities to make proper prevention and
control policies so as to effectively curb the spread of COVID-19. Different
from the 2003 SARS epidemic and the worldwide 2009 H1N1 influenza pandemic,
COVID-19 has unique epidemiological characteristics in its infectious and
recovered compartments. This drives us to formulate a new infectious dynamic
model for forecasting the COVID-19 pandemic within the human mobility network,
named the SaucIR-model in the sense that the new compartmental model extends
the benchmark SIR model by dividing the flow of people in the infected state
into asymptomatic, pathologically infected but unconfirmed, and confirmed.
Furthermore, we employ dynamic modeling of population flow in the model in
order that spatial effects can be incorporated effectively. We forecast the
spread of accumulated confirmed cases in some provinces of mainland China and
other countries that experienced severe infection during the time period from
late February to early May 2020. The novelty of incorporating the geographic
spread of the pandemic leads to a surprisingly good agreement with published
confirmed case reports. The numerical analysis validates the high degree of
predictability of our proposed SaucIR model compared to existing resemblance.
The proposed forecasting SaucIR model is implemented in Python. A web-based
application is also developed by Dash (under construction).
| stat |
MOGPTK: The Multi-Output Gaussian Process Toolkit | We present MOGPTK, a Python package for multi-channel data modelling using
Gaussian processes (GP). The aim of this toolkit is to make multi-output GP
(MOGP) models accessible to researchers, data scientists, and practitioners
alike. MOGPTK uses a Python front-end, relies on the GPflow suite and is built
on a TensorFlow back-end, thus enabling GPU-accelerated training. The toolkit
facilitates implementing the entire pipeline of GP modelling, including data
loading, parameter initialization, model learning, parameter interpretation, up
to data imputation and extrapolation. MOGPTK implements the main multi-output
covariance kernels from literature, as well as spectral-based parameter
initialization strategies. The source code, tutorials and examples in the form
of Jupyter notebooks, together with the API documentation, can be found at
http://github.com/GAMES-UChile/mogptk
| stat |
Model detection and variable selection for mode varying coefficient
model | Varying coefficient model is often used in statistical modeling since it is
more flexible than the parametric model. However, model detection and variable
selection of varying coefficient model are poorly understood in mode
regression. Existing methods in the literature for these problems often based
on mean regression and quantile regression. In this paper, we propose a novel
method to solve these problems for mode varying coefficient model based on the
B-spline approximation and SCAD penalty. Moreover, we present a new algorithm
to estimate the parameters of interest, and discuss the parameters selection
for the tuning parameters and bandwidth. We also establish the asymptotic
properties of estimated coefficients under some regular conditions. Finally, we
illustrate the proposed method by some simulation studies and an empirical
example.
| stat |
On the Minimax Optimality of the EM Algorithm for Learning Two-Component
Mixed Linear Regression | We study the convergence rates of the EM algorithm for learning two-component
mixed linear regression under all regimes of signal-to-noise ratio (SNR). We
resolve a long-standing question that many recent results have attempted to
tackle: we completely characterize the convergence behavior of EM, and show
that the EM algorithm achieves minimax optimal sample complexity under all SNR
regimes. In particular, when the SNR is sufficiently large, the EM updates
converge to the true parameter $\theta^{*}$ at the standard parametric
convergence rate $\mathcal{O}((d/n)^{1/2})$ after $\mathcal{O}(\log(n/d))$
iterations. In the regime where the SNR is above $\mathcal{O}((d/n)^{1/4})$ and
below some constant, the EM iterates converge to a $\mathcal{O}({\rm SNR}^{-1}
(d/n)^{1/2})$ neighborhood of the true parameter, when the number of iterations
is of the order $\mathcal{O}({\rm SNR}^{-2} \log(n/d))$. In the low SNR regime
where the SNR is below $\mathcal{O}((d/n)^{1/4})$, we show that EM converges to
a $\mathcal{O}((d/n)^{1/4})$ neighborhood of the true parameters, after
$\mathcal{O}((n/d)^{1/2})$ iterations. Notably, these results are achieved
under mild conditions of either random initialization or an efficiently
computable local initialization. By providing tight convergence guarantees of
the EM algorithm in middle-to-low SNR regimes, we fill the remaining gap in the
literature, and significantly, reveal that in low SNR, EM changes rate,
matching the $n^{-1/4}$ rate of the MLE, a behavior that previous work had been
unable to show.
| stat |
Knowledge Distillation as Semiparametric Inference | A popular approach to model compression is to train an inexpensive student
model to mimic the class probabilities of a highly accurate but cumbersome
teacher model. Surprisingly, this two-step knowledge distillation process often
leads to higher accuracy than training the student directly on labeled data. To
explain and enhance this phenomenon, we cast knowledge distillation as a
semiparametric inference problem with the optimal student model as the target,
the unknown Bayes class probabilities as nuisance, and the teacher
probabilities as a plug-in nuisance estimate. By adapting modern semiparametric
tools, we derive new guarantees for the prediction error of standard
distillation and develop two enhancements -- cross-fitting and loss correction
-- to mitigate the impact of teacher overfitting and underfitting on student
performance. We validate our findings empirically on both tabular and image
data and observe consistent improvements from our knowledge distillation
enhancements.
| stat |
Probabilistic Formulation of the Take The Best Heuristic | The framework of cognitively bounded rationality treats problem solving as
fundamentally rational, but emphasises that it is constrained by cognitive
architecture and the task environment. This paper investigates a simple
decision making heuristic, Take The Best (TTB), within that framework. We
formulate TTB as a likelihood-based probabilistic model, where the decision
strategy arises by probabilistic inference based on the training data and the
model constraints. The strengths of the probabilistic formulation, in addition
to providing a bounded rational account of the learning of the heuristic,
include natural extensibility with additional cognitively plausible constraints
and prior information, and the possibility to embed the heuristic as a subpart
of a larger probabilistic model. We extend the model to learn cue
discrimination thresholds for continuous-valued cues and experiment with using
the model to account for biased preference feedback from a boundedly rational
agent in a simulated interactive machine learning task.
| stat |
A test for directional-linear independence, with applications to
wildfire orientation and size | The relation between wildfire orientation and size is analyzed by means of a
nonparametric test for directional-linear independence. The test statistic is
designed for assessing the independence between two random variables of
different nature, specifically directional (fire orientation, circular or
spherical, as particular cases) and linear (fire size measured as burnt area,
scalar), based on a directional-linear nonparametric kernel density estimator.
In order to apply the proposed methodology in practice, a resampling procedure
based on permutations and bootstrap is provided. The finite sample performance
of the test is assessed by a simulation study, comparing its behavior with
other classical tests for the circular-linear case. Finally, the test is
applied to analyze wildfire data from Portugal.
| stat |
Survival Analysis via Ordinary Differential Equations | This paper introduces a general framework for survival analysis based on
ordinary differential equations (ODE). Specifically, this framework unifies
many existing survival models, including proportional hazards models, linear
transformation models, accelerated failure time models, and time-varying
coefficient models as special cases. Such a unified framework provides a novel
perspective on modeling censored data and offers opportunities for designing
new and more flexible survival model structures. Further, the aforementioned
existing survival models are traditionally estimated by procedures that suffer
from lack of scalability, statistical inefficiency, or implementation
difficulty. Based on well-established numerical solvers and sensitivity
analysis tools for ODEs, we propose a novel, scalable, and easy-to-implement
general estimation procedure that is applicable to a wide range of models. In
particular, we develop a sieve maximum likelihood estimator for a general
semi-parametric class of ODE models as an illustrative example. We also
establish a general sieve M-theorem for bundled parameters and show that the
proposed sieve estimator is consistent and asymptotically normal, and achieves
the semi-parametric efficiency bound. The finite sample performance of the
proposed estimator is examined in simulation studies and a real-world data
example.
| stat |
Clustering with the Average Silhouette Width | The Average Silhouette Width (ASW; Rousseeuw (1987)) is a popular cluster
validation index to estimate the number of clusters. Here we address the
question whether it also is suitable as a general objective function to be
optimized for finding a clustering. We will propose two algorithms (the
standard version OSil and a fast version FOSil) and compare them with existing
clustering methods in an extensive simulation study covering the cases of a
known and unknown number of clusters. Real data sets are also analysed, partly
exploring the use of the new methods with non-Euclidean distances. We will also
show that the ASW satisfies some axioms that have been proposed for cluster
quality functions (Ackerman and Ben-David (2009)). The new methods prove useful
and sensible in many cases, but some weaknesses are also highlighted. These
also concern the use of the ASW for estimating the number of clusters together
with other methods, which is of general interest due to the popularity of the
ASW for this task.
| stat |
sensobol: an R package to compute variance-based sensitivity indices | The R package "sensobol" provides several functions to conduct variance-based
uncertainty and sensitivity analysis, from the estimation of sensitivity
indices to the visual representation of the results. It implements several
state-of-the-art first and total-order estimators and allows the computation of
up to third-order effects, as well as of the approximation error, in a swift
and user-friendly way. Its flexibility makes it also appropriate for models
with either a scalar or a multivariate output. We illustrate its functionality
by conducting a variance-based sensitivity analysis of three classic models:
the Sobol' (1998) G function, the logistic population growth model of Verhulst
(1845), and the spruce budworm and forest model of Ludwig, Jones and Holling
(1976).
| stat |
Likelihood Landscape and Local Minima Structures of Gaussian Mixture
Models | In this paper, we study the landscape of the population negative
log-likelihood function of Gaussian Mixture Models with a general number of
components. Due to nonconvexity, there exist multiple local minima that are not
globally optimal, even when the mixture is well-separated. We show that all
local minima share the same form of structure that partially identifies the
component centers of the true mixture, in the sense that each local minimum
involves a non-overlapping combination of fitting multiple Gaussians to a
single true component and fitting a single Gaussian to multiple true
components. Our results apply to the setting where the true mixture components
satisfy a certain separation condition, and are valid even when the number of
components is over-or under-specified. For Gaussian mixtures with three
components, we obtain sharper results in terms of the scaling with the
separation between the components.
| stat |
Interpretable ICD Code Embeddings with Self- and Mutual-Attention
Mechanisms | We propose a novel and interpretable embedding method to represent the
international statistical classification codes of diseases and related health
problems (i.e., ICD codes). This method considers a self-attention mechanism
within the disease domain and a mutual-attention mechanism jointly between
diseases and procedures. This framework captures the clinical relationships
between the disease codes and procedures associated with hospital admissions,
and it predicts procedures according to diagnosed diseases. A self-attention
network is learned to fuse the embeddings of the diseases for each admission.
The similarities between the fused disease embedding and the procedure
embeddings indicate which procedure should potentially be recommended.
Additionally, when learning the embeddings of the ICD codes, the optimal
transport between the diseases and the procedures within each admission is
calculated as a regularizer of the embeddings. The optimal transport provides a
mutual-attention map between diseases and the procedures, which suppresses the
ambiguity within their clinical relationships. The proposed method achieves
clinically-interpretable embeddings of ICD codes, and outperforms
state-of-the-art embedding methods in procedure recommendation.
| stat |
A Python Library For Empirical Calibration | Dealing with biased data samples is a common task across many statistical
fields. In survey sampling, bias often occurs due to unrepresentative samples.
In causal studies with observational data, the treated versus untreated group
assignment is often correlated with covariates, i.e., not random. Empirical
calibration is a generic weighting method that presents a unified view on
correcting or reducing the data biases for the tasks mentioned above. We
provide a Python library EC to compute the empirical calibration weights. The
problem is formulated as convex optimization and solved efficiently in the dual
form. Compared to existing software, EC is both more efficient and robust. EC
also accommodates different optimization objectives, supports weight clipping,
and allows inexact calibration, which improves usability. We demonstrate its
usage across various experiments with both simulated and real-world data.
| stat |
The Expectation-Maximization Algorithm for Continuous-time Hidden Markov
Models | We propose a unified framework that extends the inference methods for
classical hidden Markov models to continuous settings, where both the hidden
states and observations occur in continuous time. Two different settings are
analyzed: (1) hidden jump process with a finite state space; (2) hidden
diffusion process with a continuous state space. For each setting, we first
estimate the hidden state given the observations and model parameters, showing
that the posterior distribution of the hidden states can be described by
differential equations in continuous time. Then we consider the estimation of
unknown model parameters, deriving the formulas for the
expectation-maximization algorithm in the continuous-time setting. We also
propose a Monte Carlo method for sampling the posterior distribution of the
hidden states and estimating the unknown parameters.
| stat |
Non-negative matrix and tensor factorisations with a smoothed
Wasserstein loss | Non-negative matrix and tensor factorisations are a classical tool in machine
learning and data science for finding low-dimensional representations of
high-dimensional datasets. In applications such as imaging, datasets can often
be regarded as distributions in a space with metric structure. In such a
setting, a Wasserstein loss function based on optimal transportation theory is
a natural choice since it incorporates knowledge about the geometry of the
underlying space. We introduce a general mathematical framework for computing
non-negative factorisations of matrices and tensors with respect to an optimal
transport loss, and derive an efficient method for its solution using a convex
dual formulation. We demonstrate the applicability of this approach with
several numerical examples.
| stat |
Estimating variances in time series linear regression models using
empirical BLUPs and convex optimization | We propose a two-stage estimation method of variance components in time
series models known as FDSLRMs, whose observations can be described by a linear
mixed model (LMM). We based estimating variances, fundamental quantities in a
time series forecasting approach called kriging, on the empirical (plug-in)
best linear unbiased predictions of unobservable random components in FDSLRM.
The method, providing invariant non-negative quadratic estimators, can be
used for any absolutely continuous probability distribution of time series
data. As a result of applying the convex optimization and the LMM methodology,
we resolved two problems $-$ theoretical existence and equivalence between
least squares estimators, non-negative (M)DOOLSE, and maximum likelihood
estimators, (RE)MLE, as possible starting points of our method and a practical
lack of computational implementation for FDSLRM. As for computing (RE)MLE in
the case of $ n $ observed time series values, we also discovered a new
algorithm of order $\mathcal{O}(n)$, which at the default precision is $10^7$
times more accurate and $n^2$ times faster than the best current Python(or
R)-based computational packages, namely CVXPY, CVXR, nlme, sommer and mixed.
We illustrate our results on three real data sets $-$ electricity
consumption, tourism and cyber security $-$ which are easily available,
reproducible, sharable and modifiable in the form of interactive Jupyter
notebooks.
| stat |
Anytime Online-to-Batch Conversions, Optimism, and Acceleration | A standard way to obtain convergence guarantees in stochastic convex
optimization is to run an online learning algorithm and then output the average
of its iterates: the actual iterates of the online learning algorithm do not
come with individual guarantees. We close this gap by introducing a black-box
modification to any online learning algorithm whose iterates converge to the
optimum in stochastic scenarios. We then consider the case of smooth losses,
and show that combining our approach with optimistic online learning algorithms
immediately yields a fast convergence rate of $O(L/T^{3/2}+\sigma/\sqrt{T})$ on
$L$-smooth problems with $\sigma^2$ variance in the gradients. Finally, we
provide a reduction that converts any adaptive online algorithm into one that
obtains the optimal accelerated rate of $\tilde O(L/T^2 + \sigma/\sqrt{T})$,
while still maintaining $\tilde O(1/\sqrt{T})$ convergence in the non-smooth
setting. Importantly, our algorithms adapt to $L$ and $\sigma$ automatically:
they do not need to know either to obtain these rates.
| stat |
Amortized Population Gibbs Samplers with Neural Sufficient Statistics | We develop amortized population Gibbs (APG) samplers, a class of scalable
methods that frames structured variational inference as adaptive importance
sampling. APG samplers construct high-dimensional proposals by iterating over
updates to lower-dimensional blocks of variables. We train each conditional
proposal by minimizing the inclusive KL divergence with respect to the
conditional posterior. To appropriately account for the size of the input data,
we develop a new parameterization in terms of neural sufficient statistics.
Experiments show that APG samplers can train highly structured deep generative
models in an unsupervised manner, and achieve substantial improvements in
inference accuracy relative to standard autoencoding variational methods.
| stat |
Survival analysis for AdVerse events with VarYing follow-up times
(SAVVY) -- estimation of adverse event risks | The SAVVY project aims to improve the analyses of adverse event (AE) data in
clinical trials through the use of survival techniques appropriately dealing
with varying follow-up times and competing events (CEs). Although statistical
methodologies have advanced, in AE analyses often the incidence proportion, the
incidence density, or a non-parametric Kaplan-Meier estimator (KME) are used,
which either ignore censoring or CEs. In an empirical study including
randomized clinical trials from several sponsor organisations, these potential
sources of bias are investigated. The main aim is to compare the estimators
that are typically used in AE analysis to the Aalen-Johansen estimator (AJE) as
the gold-standard. Here, one-sample findings are reported, while a companion
paper considers consequences when comparing treatment groups. Estimators are
compared with descriptive statistics, graphical displays and with a random
effects meta-analysis. The influence of different factors on the size of the
bias is investigated in a meta-regression. Comparisons are conducted at the
maximum follow-up time and at earlier evaluation time points. CEs definition
does not only include death before AE but also end of follow-up for AEs due to
events possibly related to the disease course or the treatment. Ten sponsor
organisations provided 17 trials including 186 types of AEs. The one minus KME
was on average about 1.2-fold larger than the AJE. Leading forces influencing
bias were the amount of censoring and of CEs. As a consequence, the average
bias using the incidence proportion was less than 5%. Assuming constant hazards
using incidence densities was hardly an issue provided that CEs were accounted
for. There is a need to improve the guidelines of reporting risks of AEs so
that the KME and the incidence proportion are replaced by the AJE with an
appropriate definition of CEs.
| stat |
Tight Differential Privacy for Discrete-Valued Mechanisms and for the
Subsampled Gaussian Mechanism Using FFT | We propose a numerical accountant for evaluating the tight
$(\varepsilon,\delta)$-privacy loss for algorithms with discrete one
dimensional output. The method is based on the privacy loss distribution
formalism and it uses the recently introduced Fast Fourier Transform based
accounting technique. We carry out a complete error analysis of the method in
terms of moment bounds of the privacy loss distribution which leads to rigorous
lower and upper bounds for the true $(\varepsilon,\delta)$-values. As an
application we give a novel approach to accurate privacy accounting of the
subsampled Gaussian mechanism. This completes the previously proposed analysis
by giving a strict lower and upper bounds for the
$(\varepsilon,\delta)$-values. We also demonstrate the performance of the
accountant on the binomial mechanism and show that our approach allows
decreasing noise variance up to 75 percent at equal privacy compared to
existing bounds in the literature. We also illustrate how to compute tight
bounds for the exponential mechanism applied to counting queries.
| stat |
Generalizing the Balance Heuristic Estimator in Multiple Importance
Sampling | In this paper, we propose a novel and generic family of multiple importance
sampling estimators. We first revisit the celebrated balance heuristic
estimator, a widely used Monte Carlo technique for the approximation of
intractable integrals. Then, we establish a generalized framework for the
combination of samples simulated from multiple proposals. We show that the
novel framework contains the balance heuristic as a particular case. In
addition, we study the optimal choice of the free parameters in such a way the
variance of the resulting estimator is minimized. A theoretical variance study
shows the optimal solution is always better than the balance heuristic
estimator (except in degenerate cases where both are the same). As a side
result of this analysis, we also provide new upper bounds for the balance
heuristic estimator. Finally, we show the gap in the variance of both
estimators by means of five numerical examples.
| stat |
Examining the impact of data quality and completeness of electronic
health records on predictions of patients risks of cardiovascular disease | The objective is to assess the extent of variation of data quality and
completeness of electronic health records and impact on the robustness of risk
predictions of incident cardiovascular disease (CVD) using a risk prediction
tool that is based on routinely collected data (QRISK3). The study design is a
longitudinal cohort study with a setting of 392 general practices (including
3.6 million patients) linked to hospital admission data. Variation in data
quality was assessed using Saez stability metrics quantifying outlyingness of
each practice. Statistical frailty models evaluated whether accuracy of QRISK3
predictions on individual predictions and effects of overall risk factors
(linear predictor) varied between practices. There was substantial
heterogeneity between practices in CVD incidence unaccounted for by QRISK3. In
the lowest quintile of statistical frailty, a QRISK3 predicted risk of 10% for
female was in a range between 7.1% and 9.0% when incorporating practice
variability into the statistical frailty models; for the highest quintile, this
was 10.9%-16.4%. Data quality (using Saez metrics) and completeness were
comparable across different levels of statistical frailty. For example,
recording of missing information on ethnicity was 55.7%, 62.7%, 57.8%, 64.8%
and 62.1% for practices from lowest to highest quintiles of statistical frailty
respectively. The effects of risk factors did not vary between practices with
little statistical variation of beta coefficients. In conclusion, the
considerable unmeasured heterogeneity in CVD incidence between practices was
not explained by variations in data quality or effects of risk factors. QRISK3
risk prediction should be supplemented with clinical judgement and evidence of
additional risk factors.
| stat |
Decomposition of Total Effect with the Notion of Natural Counterfactual
Interaction Effect | Mediation analysis serves as a crucial tool to obtain causal inference based
on directed acyclic graphs, which has been widely employed in the areas of
biomedical science, social science, epidemiology and psychology. Decomposition
of total effect provides a deep insight to fully understand the casual
contribution from each path and interaction term. Since the four-way
decomposition method was proposed to identify the mediated interaction effect
in counterfactual framework, the idea had been extended to a more sophisticated
scenario with non-sequential multiple mediators. However, the method exhibits
limitations as the causal structure contains direct causal edges between
mediators, such as inappropriate modeling of dependence and
non-identifiability. We develop the notion of natural counterfactual
interaction effect and find that the decomposition of total effect can be
consistently realized with our proposed notion. Furthermore, natural
counterfactual interaction effect overcomes the drawbacks and possesses a clear
and significant interpretation, which may largely improve the capacity of
researchers to analyze highly complex causal structures.
| stat |
Linear Iterative Feature Embedding: An Ensemble Framework for
Interpretable Model | A new ensemble framework for interpretable model called Linear Iterative
Feature Embedding (LIFE) has been developed to achieve high prediction
accuracy, easy interpretation and efficient computation simultaneously. The
LIFE algorithm is able to fit a wide single-hidden-layer neural network (NN)
accurately with three steps: defining the subsets of a dataset by the linear
projections of neural nodes, creating the features from multiple narrow
single-hidden-layer NNs trained on the different subsets of the data, combining
the features with a linear model. The theoretical rationale behind LIFE is also
provided by the connection to the loss ambiguity decomposition of stack
ensemble methods. Both simulation and empirical experiments confirm that LIFE
consistently outperforms directly trained single-hidden-layer NNs and also
outperforms many other benchmark models, including multi-layers Feed Forward
Neural Network (FFNN), Xgboost, and Random Forest (RF) in many experiments. As
a wide single-hidden-layer NN, LIFE is intrinsically interpretable. Meanwhile,
both variable importance and global main and interaction effects can be easily
created and visualized. In addition, the parallel nature of the base learner
building makes LIFE computationally efficient by leveraging parallel computing.
| stat |
Metric Gaussian Variational Inference | Solving Bayesian inference problems approximately with variational approaches
can provide fast and accurate results. Capturing correlation within the
approximation requires an explicit parametrization. This intrinsically limits
this approach to either moderately dimensional problems, or requiring the
strongly simplifying mean-field approach. We propose Metric Gaussian
Variational Inference (MGVI) as a method that goes beyond mean-field. Here
correlations between all model parameters are taken into account, while still
scaling linearly in computational time and memory. With this method we achieve
higher accuracy and in many cases a significant speedup compared to traditional
methods. MGVI is an iterative method that performs a series of Gaussian
approximations to the posterior. We alternate between approximating the
covariance with the inverse Fisher information metric evaluated at an
intermediate mean estimate and optimizing the KL-divergence for the given
covariance with respect to the mean. This procedure is iterated until the
uncertainty estimate is self-consistent with the mean parameter. We achieve
linear scaling by avoiding to store the covariance explicitly at any time.
Instead we draw samples from the approximating distribution relying on an
implicit representation and numerical schemes to approximately solve linear
equations. Those samples are used to approximate the KL-divergence and its
gradient. The usage of natural gradient descent allows for rapid convergence.
Formulating the Bayesian model in standardized coordinates makes MGVI
applicable to any inference problem with continuous parameters. We demonstrate
the high accuracy of MGVI by comparing it to HMC and its fast convergence
relative to other established methods in several examples. We investigate
real-data applications, as well as synthetic examples of varying size and
complexity and up to a million model parameters.
| stat |
Inferring Influence Networks from Longitudinal Bipartite Relational Data | Longitudinal bipartite relational data characterize the evolution of
relations between pairs of actors, where actors are of two distinct types and
relations exist only between disparate types. A common goal is to understand
the temporal dependencies, specifically which actor relations incite later
actor relations. There are two existing approaches to this problem. The first
approach projects the bipartite data in each time period to a unipartite
network and uses existing unipartite network models. Unfortunately, information
is lost in calculating the projection and generative models for networks
obtained through this process are scarce. The second approach represents
dependencies using two unipartite \emph{influence networks}, corresponding to
the two actor types. Existing models taking this approach are bilinear in the
influence networks, creating challenges in computation and interpretation. We
propose a novel generative model that permits estimation of weighted, directed
influence networks and does not suffer from these shortcomings. The proposed
model is linear in the influence networks, permitting inference using
off-the-shelf software tools. We prove our estimator is consistent under cases
of model misspecification and nearly asymptotically equivalent to the bilinear
estimator. We demonstrate the performance of the proposed model in simulation
studies and an analysis of weekly international state interactions.
| stat |
Learning Manifolds from Non-stationary Streaming Data | Streaming adaptations of manifold learning based dimensionality reduction
methods, such as Isomap, are based on the assumption that a small initial batch
of observations is enough for exact learning of the manifold, while remaining
streaming data instances can be cheaply mapped to this manifold. However, there
are no theoretical results to show that this core assumption is valid.
Moreover, such methods typically assume that the underlying data distribution
is stationary. Such methods are not equipped to detect, or handle, sudden
changes or gradual drifts in the distribution that may occur when the data is
streaming. We present theoretical results to show that the quality of a
manifold asymptotically converges as the size of data increases. We then show
that a Gaussian Process Regression (GPR) model, that uses a manifold-specific
kernel function and is trained on an initial batch of sufficient size, can
closely approximate the state-of-art streaming Isomap algorithms. The
predictive variance obtained from the GPR prediction is then shown to be an
effective detector of changes in the underlying data distribution. Results on
several synthetic and real data sets show that the resulting algorithm can
effectively learn lower dimensional representation of high dimensional data in
a streaming setting, while identifying shifts in the generative distribution.
| stat |
With Malice Towards None: Assessing Uncertainty via Equalized Coverage | An important factor to guarantee a fair use of data-driven recommendation
systems is that we should be able to communicate their uncertainty to decision
makers. This can be accomplished by constructing prediction intervals, which
provide an intuitive measure of the limits of predictive performance. To
support equitable treatment, we force the construction of such intervals to be
unbiased in the sense that their coverage must be equal across all protected
groups of interest. We present an operational methodology that achieves this
goal by offering rigorous distribution-free coverage guarantees holding in
finite samples. Our methodology, equalized coverage, is flexible as it can be
viewed as a wrapper around any predictive algorithm. We test the applicability
of the proposed framework on real data, demonstrating that equalized coverage
constructs unbiased prediction intervals, unlike competitive methods.
| stat |
Gradient Ascent for Active Exploration in Bandit Problems | We present a new algorithm based on an gradient ascent for a general Active
Exploration bandit problem in the fixed confidence setting. This problem
encompasses several well studied problems such that the Best Arm Identification
or Thresholding Bandits. It consists of a new sampling rule based on an online
lazy mirror ascent. We prove that this algorithm is asymptotically optimal and,
most importantly, computationally efficient.
| stat |
Calibration of the Pareto and related distributions -a
reference-intrinsic approach | We study two Bayesian (Reference Intrinsic and Jeffreys prior) and two
frequentist (MLE and PWM) approaches to calibrating the Pareto and related
distributions. Three of these approaches are compared in a simulation study and
all four to investigate how much equity risk capital banks subject to Basel II
banking regulations must hold. The Reference Intrinsic approach, which is
invariant under one-to-one transformations of the data and parameter, performs
better when fitting a generalised Pareto distribution to data simulated from a
Pareto distribution and is competitive in the case study on equity capital
requirements
| stat |
Choosing among notions of multivariate depth statistics | Classical multivariate statistics measures the outlyingness of a point by its
Mahalanobis distance from the mean, which is based on the mean and the
covariance matrix of the data. A multivariate depth function is a function
which, given a point and a distribution in d-space, measures centrality by a
number between 0 and 1, while satisfying certain postulates regarding
invariance, monotonicity, convexity and continuity. Accordingly, numerous
notions of multivariate depth have been proposed in the literature, some of
which are also robust against extremely outlying data. The departure from
classical Mahalanobis distance does not come without cost. There is a trade-off
between invariance, robustness and computational feasibility. In the last few
years, efficient exact algorithms as well as approximate ones have been
constructed and made available in R-packages. Consequently, in practical
applications the choice of a depth statistic is no more restricted to one or
two notions due to computational limits; rather often more notions are
feasible, among which the researcher has to decide. The article debates
theoretical and practical aspects of this choice, including invariance and
uniqueness, robustness and computational feasibility. Complexity and speed of
exact algorithms are compared. The accuracy of approximate approaches like the
random Tukey depth is discussed as well as the application to large and
high-dimensional data. Extensions to local and functional depths and
connections to regression depth are shortly addressed.
| stat |
Multi-Output Gaussian Processes for Crowdsourced Traffic Data Imputation | Traffic speed data imputation is a fundamental challenge for data-driven
transport analysis. In recent years, with the ubiquity of GPS-enabled devices
and the widespread use of crowdsourcing alternatives for the collection of
traffic data, transportation professionals increasingly look to such
user-generated data for many analysis, planning, and decision support
applications. However, due to the mechanics of the data collection process,
crowdsourced traffic data such as probe-vehicle data is highly prone to missing
observations, making accurate imputation crucial for the success of any
application that makes use of that type of data. In this article, we propose
the use of multi-output Gaussian processes (GPs) to model the complex spatial
and temporal patterns in crowdsourced traffic data. While the Bayesian
nonparametric formalism of GPs allows us to model observation uncertainty, the
multi-output extension based on convolution processes effectively enables us to
capture complex spatial dependencies between nearby road segments. Using 6
months of crowdsourced traffic speed data or "probe vehicle data" for several
locations in Copenhagen, the proposed approach is empirically shown to
significantly outperform popular state-of-the-art imputation methods.
| stat |
Improved Classification Based on Deep Belief Networks | For better classification generative models are used to initialize the model
and model features before training a classifier. Typically it is needed to
solve separate unsupervised and supervised learning problems. Generative
restricted Boltzmann machines and deep belief networks are widely used for
unsupervised learning. We developed several supervised models based on DBN in
order to improve this two-phase strategy. Modifying the loss function to
account for expectation with respect to the underlying generative model,
introducing weight bounds, and multi-level programming are applied in model
development. The proposed models capture both unsupervised and supervised
objectives effectively. The computational study verifies that our models
perform better than the two-phase training approach.
| stat |
Empirical bias-reducing adjustments to estimating functions | We develop a novel, general framework for reduced-bias $M$-estimation from
asymptotically unbiased estimating functions. The framework relies on an
empirical approximation of the bias by a function of derivatives of estimating
function contributions. Reduced-bias $M$-estimation operates either implicitly,
by solving empirically adjusted estimating equations, or explicitly, by
subtracting the estimated bias from the original $M$-estimates, and applies to
models that are partially- or fully-specified, with either likelihoods or other
surrogate objectives. Automatic differentiation can be used to abstract away
the only algebra required to implement reduced-bias $M$-estimation. As a
result, the bias reduction methods we introduce have markedly broader
applicability and more straightforward implementation than other established
bias-reduction methods that require resampling or evaluation of expectations of
products of log-likelihood derivatives. If $M$-estimation is by maximizing an
objective, then there always exists a bias-reducing penalized objective. That
penalized objective relates closely to information criteria for model
selection, and can be further enhanced with plug-in penalties to deliver
reduced-bias $M$-estimates with extra properties, like finiteness in models for
categorical data. The reduced-bias $M$-estimators have the same asymptotic
distribution as the original $M$-estimators, and, hence, standard procedures
for inference and model selection apply unaltered with the improved estimates.
We demonstrate and assess the properties of reduced-bias $M$-estimation in
well-used, prominent modelling settings of varying complexity.
| stat |
Self-Validated Ensemble Models for Design of Experiments | In this paper we introduce a new model building algorithm called
self-validating ensemble modeling or SVEM. The method enables the fitting of
validated predictive models to the relatively small data sets typically
generated from designed experiments where prediction is the desired outcome
which is often the case in Quality by Design studies in bio-pharmaceutical
industries. In order to fit validated predictive models, SVEM uses a unique
weighting scheme applied to the responses and fractional weighted bootstrapping
to generate a large ensemble of fitted models. The weighting scheme allows the
original data to serve both as a training set and validation set. The method is
very general in application and works with most model selection algorithms.
Through extensive simulation studies and a case study we demonstrate that SVEM
generates models with lower prediction error as compared to more traditional
statistical approaches that are based on fitting a single model.
| stat |
Neural Connectivity with Hidden Gaussian Graphical State-Model | The noninvasive procedures for neural connectivity are under questioning.
Theoretical models sustain that the electromagnetic field registered at
external sensors is elicited by currents at neural space. Nevertheless, what we
observe at the sensor space is a superposition of projected fields, from the
whole gray-matter. This is the reason for a major pitfall of noninvasive
Electrophysiology methods: distorted reconstruction of neural activity and its
connectivity or leakage. It has been proven that current methods produce
incorrect connectomes. Somewhat related to the incorrect connectivity
modelling, they disregard either Systems Theory and Bayesian Information
Theory. We introduce a new formalism that attains for it, Hidden Gaussian
Graphical State-Model (HIGGS). A neural Gaussian Graphical Model (GGM) hidden
by the observation equation of Magneto-encephalographic (MEEG) signals. HIGGS
is equivalent to a frequency domain Linear State Space Model (LSSM) but with
sparse connectivity prior. The mathematical contribution here is the theory for
high-dimensional and frequency-domain HIGGS solvers. We demonstrate that HIGGS
can attenuate the leakage effect in the most critical case: the distortion EEG
signal due to head volume conduction heterogeneities. Its application in EEG is
illustrated with retrieved connectivity patterns from human Steady State Visual
Evoked Potentials (SSVEP). We provide for the first time confirmatory evidence
for noninvasive procedures of neural connectivity: concurrent EEG and
Electrocorticography (ECoG) recordings on monkey. Open source packages are
freely available online, to reproduce the results presented in this paper and
to analyze external MEEG databases.
| stat |
Overcoming the inconsistences of the variance inflation factor: a
redefined VIF and a test to detect statistical troubling multicollinearity | Multicollinearity is relevant to many different fields where linear
regression models are applied, and its existence may affect the analysis of
ordinary least squares (OLS) estimators from both the numerical and statistical
points of views. Thus, multicollinearity can lead to incoherence in the
statistical significance of the independent variables and the global
significance of the model. The variance inflation factor (VIF) is traditionally
applied to diagnose the possible existence of multicollinearity, but it is not
always the case that detection by VIF of a troubling degree of
multicollinearity corresponds to negative effects on the statistical analysis.
The reason for the lack of specificity of VIF is that there are other factors,
such as the size of the sample and the variance of the random disturbance, that
can lead to high values of the VIF but not to problematic variance in the OLS
estimators (see O'Brien 2007). This paper presents a new variance inflation
factor (TVIF) that consider all these additional factors. Thresholds for this
new measure and from the index provided by Stewart (1987) are also provided.
These thresholds are reinterpreted and presented as a new statistical test to
diagnose the existence of statistical troubling multicollinearity. The
contributions of this paper are illustrated with two real data examples
previously applied in the scientific literature.
| stat |
Selective inference after feature selection via multiscale bootstrap | It is common to show the confidence intervals or $p$-values of selected
features, or predictor variables in regression, but they often involve
selection bias. The selective inference approach solves this bias by
conditioning on the selection event. Most existing studies of selective
inference consider a specific algorithm, such as Lasso, for feature selection,
and thus they have difficulties in handling more complicated algorithms.
Moreover, existing studies often consider unnecessarily restrictive events,
leading to over-conditioning and lower statistical power. Our novel and
widely-applicable resampling method addresses these issues to compute an
approximately unbiased selective $p$-value for the selected features. We prove
that the $p$-value computed by our resampling method is more accurate and more
powerful than existing methods, while the computational cost is the same order
as the classical bootstrap method. Numerical experiments demonstrate that our
algorithm works well even for more complicated feature selection methods such
as non-convex regularization.
| stat |
Moment-Based Domain Adaptation: Learning Bounds and Algorithms | This thesis contributes to the mathematical foundation of domain adaptation
as emerging field in machine learning. In contrast to classical statistical
learning, the framework of domain adaptation takes into account deviations
between probability distributions in the training and application setting.
Domain adaptation applies for a wider range of applications as future samples
often follow a distribution that differs from the ones of the training samples.
A decisive point is the generality of the assumptions about the similarity of
the distributions. Therefore, in this thesis we study domain adaptation
problems under as weak similarity assumptions as can be modelled by finitely
many moments.
| stat |
The wrapped xgamma distribution for modeling circular data appearing in
geological context | The technique of wrapping of a univariate probability distribution is very
effective in getting a circular form of the underlying density. In this
article, we introduce the circular (wrapped) version of xgamma distribution and
study its different distributional properties. To estimate the unknown
parameter, maximum likelihood method is proposed. A Monte-Carlo simulation
study is performed to understand the behaviour of the estimates for varying
sample size. To illustrate the application of the proposed distribution, a real
data set on the long axis orientation of feldspar laths in basalt rock is
analyzed and compared with other circular distributions.
| stat |
Augmenting expert detection of early coronary artery occlusion from 12
lead electrocardiograms using deep learning | Early diagnosis of acute coronary artery occlusion based on electrocardiogram
(ECG) findings is essential for prompt delivery of primary percutaneous
coronary intervention. Current ST elevation (STE) criteria are specific but
insensitive. Consequently, it is likely that many patients are missing out on
potentially life-saving treatment. Experts combining non-specific ECG changes
with STE detect ischaemia with higher sensitivity, but at the cost of
specificity. We show that a deep learning model can detect ischaemia caused by
acute coronary artery occlusion with a better balance of sensitivity and
specificity than STE criteria, existing computerised analysers or expert
cardiologists.
| stat |
Estimating Individualized Treatment Regimes from Crossover Designs | The field of precision medicine aims to tailor treatment based on
patient-specific factors in a reproducible way. To this end, estimating an
optimal individualized treatment regime (ITR) that recommends treatment
decisions based on patient characteristics to maximize the mean of a
pre-specified outcome is of particular interest. Several methods have been
proposed for estimating an optimal ITR from clinical trial data in the parallel
group setting where each subject is randomized to a single intervention.
However, little work has been done in the area of estimating the optimal ITR
from crossover study designs. Such designs naturally lend themselves to
precision medicine, because they allow for observing the response to multiple
treatments for each patient. In this paper, we introduce a method for
estimating the optimal ITR using data from a 2x2 crossover study with or
without carryover effects. The proposed method is similar to policy search
methods such as outcome weighted learning; however, we take advantage of the
crossover design by using the difference in responses under each treatment as
the observed reward. We establish Fisher and global consistency, present
numerical experiments, and analyze data from a feeding trial to demonstrate the
improved performance of the proposed method compared to standard methods for a
parallel study design.
| stat |
Hydra: A method for strain-minimizing hyperbolic embedding of network-
and distance-based data | We introduce hydra (hyperbolic distance recovery and approximation), a new
method for embedding network- or distance-based data into hyperbolic space. We
show mathematically that hydra satisfies a certain optimality guarantee: It
minimizes the `hyperbolic strain' between original and embedded data points.
Moreover, it recovers points exactly, when they are located on a hyperbolic
submanifold of the feature space. Testing on real network data we show that the
embedding quality of hydra is competitive with existing hyperbolic embedding
methods, but achieved at substantially shorter computation time. An extended
method, termed hydra+, outperforms existing methods in both computation time
and embedding quality.
| stat |
Uniformly Valid Post-Regularization Confidence Regions for Many
Functional Parameters in Z-Estimation Framework | In this paper we develop procedures to construct simultaneous confidence
bands for $\tilde p$ potentially infinite-dimensional parameters after model
selection for general moment condition models where $\tilde p$ is potentially
much larger than the sample size of available data, $n$. This allows us to
cover settings with functional response data where each of the $\tilde p$
parameters is a function. The procedure is based on the construction of score
functions that satisfy certain orthogonality condition. The proposed
simultaneous confidence bands rely on uniform central limit theorems for
high-dimensional vectors (and not on Donsker arguments as we allow for $\tilde
p \gg n$). To construct the bands, we employ a multiplier bootstrap procedure
which is computationally efficient as it only involves resampling the estimated
score functions (and does not require resolving the high-dimensional
optimization problems). We formally apply the general theory to inference on
regression coefficient process in the distribution regression model with a
logistic link, where two implementations are analyzed in detail. Simulations
and an application to real data are provided to help illustrate the
applicability of the results.
| stat |
Optimal Clustering with Missing Values | Missing values frequently arise in modern biomedical studies due to various
reasons, including missing tests or complex profiling technologies for
different omics measurements. Missing values can complicate the application of
clustering algorithms, whose goals are to group points based on some similarity
criterion. A common practice for dealing with missing values in the context of
clustering is to first impute the missing values, and then apply the clustering
algorithm on the completed data.
We consider missing values in the context of optimal clustering, which finds
an optimal clustering operator with reference to an underlying random labeled
point process (RLPP). We show how the missing-value problem fits neatly into
the overall framework of optimal clustering by incorporating the missing value
mechanism into the random labeled point process and then marginalizing out the
missing-value process. In particular, we demonstrate the proposed framework for
the Gaussian model with arbitrary covariance structures. Comprehensive
experimental studies on both synthetic and real-world RNA-seq data show the
superior performance of the proposed optimal clustering with missing values
when compared to various clustering approaches. Optimal clustering with missing
values obviates the need for imputation-based pre-processing of the data, while
at the same time possessing smaller clustering errors.
| stat |
Long-Term Effect Estimation with Surrogate Representation | There are many scenarios where short- and long-term causal effects of an
intervention are different. For example, low-quality ads may increase
short-term ad clicks but decrease the long-term revenue via reduced clicks.
This work, therefore, studies the problem of long-term effect where the outcome
of primary interest, or primary outcome, takes months or even years to
accumulate. The observational study of long-term effect presents unique
challenges. First, the confounding bias causes large estimation error and
variance, which can further accumulate towards the prediction of primary
outcomes. Second, short-term outcomes are often directly used as the proxy of
the primary outcome, i.e., the surrogate. Nevertheless, this method entails the
strong surrogacy assumption that is often impractical. To tackle these
challenges, we propose to build connections between long-term causal inference
and sequential models in machine learning. This enables us to learn surrogate
representations that account for the temporal unconfoundedness and circumvent
the stringent surrogacy assumption by conditioning on the inferred time-varying
confounders. Experimental results show that the proposed framework outperforms
the state-of-the-art.
| stat |
Big Data and model-based survey sampling | Big Data are huge amounts of digital information that are automatically
accrued or merged from several sources and rarely result from properly planned
surveys. A Big Dataset is herein conceived of as a collection of information
concerning a finite population. We suggest selecting a sample of observations
to get the inferential goal. We assume a super-population model has generated
the Big Dataset. With this assumption, we can apply the theory of optimal
design to draw a sample from the Big Dataset that contains the majority of the
information about the unknown parameters.
| stat |
When and How to Lift the Lockdown? Global COVID-19 Scenario Analysis and
Policy Assessment using Compartmental Gaussian Processes | The coronavirus disease 2019 (COVID-19) global pandemic has led many
countries to impose unprecedented lockdown measures in order to slow down the
outbreak. Questions on whether governments have acted promptly enough, and
whether lockdown measures can be lifted soon have since been central in public
discourse. Data-driven models that predict COVID-19 fatalities under different
lockdown policy scenarios are essential for addressing these questions and
informing governments on future policy directions. To this end, this paper
develops a Bayesian model for predicting the effects of COVID-19 lockdown
policies in a global context -- we treat each country as a distinct data point,
and exploit variations of policies across countries to learn country-specific
policy effects. Our model utilizes a two-layer Gaussian process (GP) prior --
the lower layer uses a compartmental SEIR (Susceptible, Exposed, Infected,
Recovered) model as a prior mean function with "country-and-policy-specific"
parameters that capture fatality curves under "counterfactual" policies within
each country, whereas the upper layer is shared across all countries, and
learns lower-layer SEIR parameters as a function of a country's features and
its policy indicators. Our model combines the solid mechanistic foundations of
SEIR models (Bayesian priors) with the flexible data-driven modeling and
gradient-based optimization routines of machine learning (Bayesian posteriors)
-- i.e., the entire model is trained end-to-end via stochastic variational
inference. We compare the projections of COVID-19 fatalities by our model with
other models listed by the Center for Disease Control (CDC), and provide
scenario analyses for various lockdown and reopening strategies highlighting
their impact on COVID-19 fatalities.
| stat |
ARIMA forecasting of COVID-19 incidence in Italy, Russia, and the USA | The novel Coronavirus disease (COVID-19) is a severe respiratory infection
that officially occurred in Wuhan, China, in December 2019. In late February,
the disease began to spread quickly across the world, causing serious health,
social, and economic emergencies. This paper aims to forecast the short to
medium-term incidence of COVID-19 epidemic through the medium of an
autoregressive integrated moving average (ARIMA) model, applied to Italy,
Russia, and the USA The analysis is carried out on the number of new daily
confirmed COVID-19 cases, collected by Worldometer website. The best ARIMA
models are Italy (4,2,4), Russia (1,2,1), and the USA (6,2,3). The results show
that: i) ARIMA models are reliable enough when new daily cases begin to
stabilize; ii) Italy, the USA, and Russia reached the peak of COVID-19
infections in mid-April, mid-May, and late May, respectively; and iii) Russia
and the USA will require much more time than Italy to drop COVID-19 cases near
zero. This may suggest the importance of the application of quick and effective
lockdown measures, which have been relatively stricter in Italy. Therefore,
even if the results should be interpreted with caution, ARIMA models seem to be
a good tool that can help the health authorities to monitor the diffusion of
the outbreak.
| stat |
Distributed estimation of principal support vector machines for
sufficient dimension reduction | The principal support vector machines method (Li et al., 2011) is a powerful
tool for sufficient dimension reduction that replaces original predictors with
their low-dimensional linear combinations without loss of information. However,
the computational burden of the principal support vector machines method
constrains its use for massive data. To address this issue, we in this paper
propose two distributed estimation algorithms for fast implementation when the
sample size is large. Both the two distributed sufficient dimension reduction
estimators enjoy the same statistical efficiency as merging all the data
together, which provides rigorous statistical guarantees for their application
to large scale datasets. The two distributed algorithms are further adapt to
principal weighted support vector machines (Shin et al., 2016) for sufficient
dimension reduction in binary classification. The statistical accuracy and
computational complexity of our proposed methods are examined through
comprehensive simulation studies and a real data application with more than
600000 samples.
| stat |
To Aid Scientific Inference, Emphasize Unconditional Descriptions of
Statistics | We have elsewhere reviewed proposals to reform terminology and improve
interpretations of conventional statistics by emphasizing logical and
information concepts over probability concepts. We here give detailed reasons
and methods for reinterpreting statistics (including but not limited to)
P-values and interval estimates in unconditional terms, which describe
compatibility of observations with an entire set of analysis assumptions,
rather than just a narrow target hypothesis. Such reinterpretations help avoid
overconfident inferences whenever there is uncertainty about the assumptions
used to derive and compute the statistical results. Examples of such
assumptions include not only standard statistical modeling assumptions, but
also assumptions about absence of systematic errors, protocol violations, and
data corruption. Unconditional descriptions introduce uncertainty about such
assumptions directly into statistical presentations of results, rather than
leaving that only to the informal discussion that ensues. We thus view
unconditional description as a vital component of good statistical training and
presentation.
| stat |
Boltzmann Generators -- Sampling Equilibrium States of Many-Body Systems
with Deep Learning | Computing equilibrium states in condensed-matter many-body systems, such as
solvated proteins, is a long-standing challenge. Lacking methods for generating
statistically independent equilibrium samples in "one shot", vast computational
effort is invested for simulating these system in small steps, e.g., using
Molecular Dynamics. Combining deep learning and statistical mechanics, we here
develop Boltzmann Generators, that are shown to generate unbiased one-shot
equilibrium samples of representative condensed matter systems and proteins.
Boltzmann Generators use neural networks to learn a coordinate transformation
of the complex configurational equilibrium distribution to a distribution that
can be easily sampled. Accurate computation of free energy differences and
discovery of new configurations are demonstrated, providing a statistical
mechanics tool that can avoid rare events during sampling without prior
knowledge of reaction coordinates.
| stat |
Optimal designs for some bivariate cokriging models | This article focuses on the estimation and design aspects of a bivariate
collocated cokriging experiment. For a large class of covariance matrices a
linear dependency criterion is identified, which allows the best linear
unbiased estimator of the primary variable in a bivariate collocated cokriging
setup to reduce to a univariate kriging estimator. Exact optimal designs for
efficient prediction for such simple and ordinary cokriging models, with one
dimensional inputs are determined. Designs are found by minimizing the maximum
and integrated prediction variance. For simple and ordinary cokriging models
with known covariance parameters, the equispaced design is shown to be optimal
for both criterion functions. The more realistic scenario of unknown covariance
parameters is addressed by assuming prior distributions on the parameter
vector, thus adopting a Bayesian approach to the design problem. The equispaced
design is proved to be the Bayesian optimal design for both criteria. The work
is motivated by designing an optimal water monitoring system for an Indian
river.
| stat |
Geometric Losses for Distributional Learning | Building upon recent advances in entropy-regularized optimal transport, and
upon Fenchel duality between measures and continuous functions , we propose a
generalization of the logistic loss that incorporates a metric or cost between
classes. Unlike previous attempts to use optimal transport distances for
learning, our loss results in unconstrained convex objective functions,
supports infinite (or very large) class spaces, and naturally defines a
geometric generalization of the softmax operator. The geometric properties of
this loss make it suitable for predicting sparse and singular distributions,
for instance supported on curves or hyper-surfaces. We study the theoretical
properties of our loss and show-case its effectiveness on two applications:
ordinal regression and drawing generation.
| stat |
Privacy-preserving Q-Learning with Functional Noise in Continuous State
Spaces | We consider differentially private algorithms for reinforcement learning in
continuous spaces, such that neighboring reward functions are
indistinguishable. This protects the reward information from being exploited by
methods such as inverse reinforcement learning. Existing studies that guarantee
differential privacy are not extendable to infinite state spaces, as the noise
level to ensure privacy will scale accordingly to infinity. Our aim is to
protect the value function approximator, without regard to the number of states
queried to the function. It is achieved by adding functional noise to the value
function iteratively in the training. We show rigorous privacy guarantees by a
series of analyses on the kernel of the noise space, the probabilistic bound of
such noise samples, and the composition over the iterations. We gain insight
into the utility analysis by proving the algorithm's approximate optimality
when the state space is discrete. Experiments corroborate our theoretical
findings and show improvement over existing approaches.
| stat |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.