title
stringlengths 6
244
| abstract
stringlengths 19
5.09k
| label
stringclasses 10
values |
|---|---|---|
PRISM: Patient Response Identifiers for Stratified Medicine | Pharmaceutical companies continue to seek innovative ways to explore whether
a drug under development is likely to be suitable for all or only an
identifiable stratum of patients in the target population. The sooner this can
be done during the clinical development process, the better it is for the
company, and downstream for prescribers, payers, and most importantly, for
patients. To help enable this vision of stratified medicine, we describe a
powerful statistical framework, Patient Response Identifiers for Stratified
Medicine (PRISM), for the discovery of potential predictors of drug response
and associated subgroups using machine learning tools. PRISM is highly flexible
and can have many "configurations", allowing the incorporation of complementary
models or tools for a variety of outcomes and settings. One promising PRISM
configuration is to use the observed outcomes for subgroup identification,
while using counterfactual within-patient predicted treatment differences for
subgroup-specific treatment estimates and associated interpretation. This
separates the "subgroup-identification" from the "decision-making" and, to
facilitate clinical design planning, is a simple way to obtain unbiased
treatment effect sizes in the discovered subgroups. Simulation results, along
with data from a real clinical trial are used to illustrate the utility of the
proposed PRISM framework.
| stat |
Modeling Bivariate Geyser Eruption System with Covariate-Adjusted
Recurrent Event Process | Geyser eruption is one of the most popular signature attractions at the
Yellowstone National Park. The interdependence of geyser eruptions and impacts
of covariates are of interest to researchers in geyser studies. In this paper,
we propose a parametric covariate-adjusted recurrent event model for estimating
the eruption gap time. We describe a general bivariate recurrent event process,
where a bivariate lognormal distribution and a Gumbel copula with different
marginal distributions are used to model an interdependent dual-type event
system. The maximum likelihood approach is used to estimate model parameters.
The proposed method is applied to analyzing the Yellowstone geyser eruption
data for a bivariate geyser system and offers a deeper understanding of the
event occurrence mechanism of individual events as well as the system as a
whole. A comprehensive simulation study is conducted to evaluate the
performance of the proposed method.
| stat |
Spatial Modeling of Heavy Precipitation by Coupling Weather Station
Recordings and Ensemble Forecasts with Max-Stable Processes | Due to complex physical phenomena, the distribution of heavy rainfall events
is difficult to model spatially. Physically based numerical models can often
provide physically coherent spatial patterns, but may miss some important
precipitation features like heavy rainfall intensities. Measurements at
ground-based weather stations, however, supply adequate rainfall intensities,
but most national weather recording networks are often spatially too sparse to
capture rainfall patterns adequately. To bring the best out of these two
sources of information, climatologists and hydrologists have been seeking
models that can efficiently merge different types of rainfall data. One
inherent difficulty is to capture the appropriate multivariate dependence
structure among rainfall maxima. For this purpose, multivariate extreme value
theory suggests the use of a max-stable process. Such a process can be
represented by a max-linear combination of independent copies of a hidden
stochastic process weighted by a Poisson point process. In practice, the choice
of this hidden process is non-trivial, especially if anisotropy,
non-stationarity and nugget effects are present in the spatial data at hand. By
coupling forecast ensemble data from the French national weather service
(M\'et\'eo-France) with local observations, we construct and compare different
types of data driven max-stable processes that are parsimonious in parameters,
easy to simulate and capable of reproducing nugget effects and spatial
non-stationarities. We also compare our new method with classical approaches
from spatial extreme value theory such as Brown-Resnick processes.
| stat |
Bayesian Analysis of Glucose Dynamics during the Oral Glucose Tolerance
Test (OGTT) | In this paper, we propose a model of the dynamics of the blood glucose level
during an Oral Glucose Tolerance Test (OGTT). This dynamic includes the action
of insulin and glucagon in the glucose homeostasis process as the reaction of
an oral stimulus. We propose a Bayesian approach in the inference of five
parameters related to insulin secretion, glucagon secretion, gastrointestinal
emptying, and basal glucose level. Two insulin indicators related to the
glucose level in blood and in the gastrointestinal tract allow us to suggest a
classification for patients with impaired insulin sensitivity.
| stat |
Quasi-Bayesian Inference for Production Frontiers | We propose a quasi-Bayesian method to conduct inference for the production
frontier. This approach combines multiple first-stage extreme quantile
estimates by the quasi-Bayesian method to produce the point estimate and
confidence interval for the production frontier. We show the asymptotic
properties of the proposed estimator and the validity of the inference
procedure. The finite sample performance of our method is illustrated through
simulations and an empirical application.
| stat |
Machine Learning for Set-Identified Linear Models | This paper provides estimation and inference methods for an identified set
where the selection among a very large number of covariates is based on modern
machine learning tools. I characterize the boundary of the identified set
(i.e., support function) using a semiparametric moment condition. Combining
Neyman-orthogonality and sample splitting ideas, I construct a root-N
consistent, uniformly asymptotically Gaussian estimator of the support function
and propose a weighted bootstrap procedure to conduct inference about the
identified set. I provide a general method to construct a Neyman-orthogonal
moment condition for the support function. Applying my method to Lee (2008)'s
endogenous selection model, I provide the asymptotic theory for the sharp
(i.e., the tightest possible) bounds on the Average Treatment Effect in the
presence of high-dimensional covariates. Furthermore, I relax the conventional
monotonicity assumption and allow the sign of the treatment effect on the
selection (e.g., employment) to be determined by covariates. Using JobCorps
data set with very rich baseline characteristics, I substantially tighten the
bounds on the JobCorps effect on wages under weakened monotonicity assumption.
| stat |
Who Learns Better Bayesian Network Structures: Accuracy and Speed of
Structure Learning Algorithms | Three classes of algorithms to learn the structure of Bayesian networks from
data are common in the literature: constraint-based algorithms, which use
conditional independence tests to learn the dependence structure of the data;
score-based algorithms, which use goodness-of-fit scores as objective functions
to maximise; and hybrid algorithms that combine both approaches.
Constraint-based and score-based algorithms have been shown to learn the same
structures when conditional independence and goodness of fit are both assessed
using entropy and the topological ordering of the network is known (Cowell,
2001).
In this paper, we investigate how these three classes of algorithms perform
outside the assumptions above in terms of speed and accuracy of network
reconstruction for both discrete and Gaussian Bayesian networks. We approach
this question by recognising that structure learning is defined by the
combination of a statistical criterion and an algorithm that determines how the
criterion is applied to the data. Removing the confounding effect of different
choices for the statistical criterion, we find using both simulated and
real-world complex data that constraint-based algorithms are often less
accurate than score-based algorithms, but are seldom faster (even at large
sample sizes); and that hybrid algorithms are neither faster nor more accurate
than constraint-based algorithms. This suggests that commonly held beliefs on
structure learning in the literature are strongly influenced by the choice of
particular statistical criteria rather than just by the properties of the
algorithms themselves.
| stat |
High-frequency Estimation of the L\'evy-driven Graph Ornstein-Uhlenbeck
process | We consider the Graph Ornstein-Uhlenbeck (GrOU) process observed on a
non-uniform discrete time grid and introduce discretised maximum likelihood
estimators with parameters specific to the whole graph or specific to each
component, or node. Under a high-frequency sampling scheme, we study the
asymptotic behaviour of those estimators as the mesh size of the observation
grid goes to zero. We prove two stable central limit theorems to the same
distribution as in the continuously-observed case under both finite and
infinite jump activity for the L\'evy driving noise. When a graph structure is
not explicitly available, the stable convergence allows to consider
purpose-specific sparse inference procedures, i.e. pruning, on the edges
themselves in parallel to the GrOU inference and preserve its asymptotic
properties. We apply the new estimators to wind capacity factor measurements,
i.e. the ratio between the wind power produced locally compared to its rated
peak power, across fifty locations in Northern Spain and Portugal. We show the
superiority of those estimators compared to the standard least squares
estimator through a simulation study extending known univariate results across
graph configurations, noise types and amplitudes.
| stat |
Fast implementation of partial least squares for function-on-function
regression | People employ the function-on-function regression to model the relationship
between two random curves. Fitting this model, widely used strategies include
algorithms falling into the framework of functional partial least squares
(typically requiring iterative eigen-decomposition). Here we introduce a route
of functional partial least squares based upon Krylov subspaces. It can be
expressed in two forms equivalent to each other (in exact arithmetic): one is
non-iterative with explicit forms of estimators and predictions, facilitating
the theoretical derivation and potential extensions (to more complex models);
the other one stabilizes numerical outputs. The consistence of estimators and
predictions is established under regularity conditions. Our proposal is
highlighted as it is less computationally involved. Meanwhile, it is
competitive in terms of both estimation and prediction accuracy.
| stat |
Penalized estimation of flexible hidden Markov models for time series of
counts | Hidden Markov models are versatile tools for modeling sequential
observations, where it is assumed that a hidden state process selects which of
finitely many distributions generates any given observation. Specifically for
time series of counts, the Poisson family often provides a natural choice for
the state-dependent distributions, though more flexible distributions such as
the negative binomial or distributions with a bounded range can also be used.
However, in practice, choosing an adequate class of (parametric) distributions
is often anything but straightforward, and an inadequate choice can have severe
negative consequences on the model's predictive performance, on state
classification, and generally on inference related to the system considered. To
address this issue, we propose an effectively nonparametric approach to fitting
hidden Markov models to time series of counts, where the state-dependent
distributions are estimated in a completely data-driven way without the need to
select a distributional family. To avoid overfitting, we add a roughness
penalty based on higher-order differences between adjacent count probabilities
to the likelihood, which is demonstrated to produce smooth probability mass
functions of the state-dependent distributions. The feasibility of the
suggested approach is assessed in a simulation experiment, and illustrated in
two real-data applications, where we model the distribution of i) major
earthquake counts and ii) acceleration counts of an oceanic whitetip shark
(Carcharhinus longimanus) over time.
| stat |
Do Deep Generative Models Know What They Don't Know? | A neural network deployed in the wild may be asked to make predictions for
inputs that were drawn from a different distribution than that of the training
data. A plethora of work has demonstrated that it is easy to find or synthesize
inputs for which a neural network is highly confident yet wrong. Generative
models are widely viewed to be robust to such mistaken confidence as modeling
the density of the input features can be used to detect novel,
out-of-distribution inputs. In this paper we challenge this assumption. We find
that the density learned by flow-based models, VAEs, and PixelCNNs cannot
distinguish images of common objects such as dogs, trucks, and horses (i.e.
CIFAR-10) from those of house numbers (i.e. SVHN), assigning a higher
likelihood to the latter when the model is trained on the former. Moreover, we
find evidence of this phenomenon when pairing several popular image data sets:
FashionMNIST vs MNIST, CelebA vs SVHN, ImageNet vs CIFAR-10 / CIFAR-100 / SVHN.
To investigate this curious behavior, we focus analysis on flow-based
generative models in particular since they are trained and evaluated via the
exact marginal likelihood. We find such behavior persists even when we restrict
the flows to constant-volume transformations. These transformations admit some
theoretical analysis, and we show that the difference in likelihoods can be
explained by the location and variances of the data and the model curvature.
Our results caution against using the density estimates from deep generative
models to identify inputs similar to the training distribution until their
behavior for out-of-distribution inputs is better understood.
| stat |
Optimal Estimation of Generalized Average Treatment Effects using Kernel
Optimal Matching | In causal inference, a variety of causal effect estimands have been studied,
including the sample, uncensored, target, conditional, optimal subpopulation,
and optimal weighted average treatment effects. Ad-hoc methods have been
developed for each estimand based on inverse probability weighting (IPW) and on
outcome regression modeling, but these may be sensitive to model
misspecification, practical violations of positivity, or both. The contribution
of this paper is twofold. First, we formulate the generalized average treatment
effect (GATE) to unify these causal estimands as well as their IPW estimates.
Second, we develop a method based on Kernel Optimal Matching (KOM) to optimally
estimate GATE and to find the GATE most easily estimable by KOM, which we term
the Kernel Optimal Weighted Average Treatment Effect. KOM provides uniform
control on the conditional mean squared error of a weighted estimator over a
class of models while simultaneously controlling for precision. We study its
theoretical properties and evaluate its comparative performance in a simulation
study. We illustrate the use of KOM for GATE estimation in two case studies:
comparing spine surgical interventions and studying the effect of peer support
on people living with HIV.
| stat |
A General Class of New Continuous Mixture Distribution and Application | A generalization of a distribution increases the flexibility particularly in
studying of a phenomenon and its properties. Many generalizations of continuous
univariate distributions are available in literature. In this study, an
investigation is conducted on a distribution and its generalization. Several
available generalizations of the distribution are reviewed and recent trends in
the construction of generalized classes with a generalized mixing parameter are
discussed. To check the suitability and comparability, real data set have been
used.
| stat |
A mixture model approach for clustering bipartite networks | This chapter investigates the latent structure of bipartite networks via a
model-based clustering approach which is able to capture both latent groups of
sending nodes and latent variability of the propensity of sending nodes to
create links with receiving nodes within each group. This modelling approach is
very flexible and can be estimated by using fast inferential approaches such as
variational inference. We apply this model to the analysis of a terrorist
network in order to identify the main latent groups of terrorists and their
latent trait scores based on their attendance to some events.
| stat |
Quantifying Statistical Significance of Neural Network
Representation-Driven Hypotheses by Selective Inference | In the past few years, various approaches have been developed to explain and
interpret deep neural network (DNN) representations, but it has been pointed
out that these representations are sometimes unstable and not reproducible. In
this paper, we interpret these representations as hypotheses driven by DNN
(called DNN-driven hypotheses) and propose a method to quantify the reliability
of these hypotheses in statistical hypothesis testing framework. To this end,
we introduce Selective Inference (SI) framework, which has received much
attention in the past few years as a new statistical inference framework for
data-driven hypotheses. The basic idea of SI is to make conditional inferences
on the selected hypotheses under the condition that they are selected. In order
to use SI framework for DNN representations, we develop a new SI algorithm
based on homotopy method which enables us to derive the exact (non-asymptotic)
conditional sampling distribution of the DNN-driven hypotheses. We conduct
experiments on both synthetic and real-world datasets, through which we offer
evidence that our proposed method can successfully control the false positive
rate, has decent performance in terms of computational efficiency, and provides
good results in practical applications.
| stat |
Approximate Gradient Coding with Optimal Decoding | In distributed optimization problems, a technique called gradient coding,
which involves replicating data points, has been used to mitigate the effect of
straggling machines. Recent work has studied approximate gradient coding, which
concerns coding schemes where the replication factor of the data is too low to
recover the full gradient exactly. Our work is motivated by the challenge of
creating approximate gradient coding schemes that simultaneously work well in
both the adversarial and stochastic models. To that end, we introduce novel
approximate gradient codes based on expander graphs, in which each machine
receives exactly two blocks of data points. We analyze the decoding error both
in the random and adversarial straggler setting, when optimal decoding
coefficients are used. We show that in the random setting, our schemes achieve
an error to the gradient that decays exponentially in the replication factor.
In the adversarial setting, the error is nearly a factor of two smaller than
any existing code with similar performance in the random setting. We show
convergence bounds both in the random and adversarial setting for gradient
descent under standard assumptions using our codes. In the random setting, our
convergence rate improves upon block-box bounds. In the adversarial setting, we
show that gradient descent can converge down to a noise floor that scales
linearly with the adversarial error to the gradient. We demonstrate empirically
that our schemes achieve near-optimal error in the random setting and converge
faster than algorithms which do not use the optimal decoding coefficients.
| stat |
Endemic-epidemic models with discrete-time serial interval distributions
for infectious disease prediction | Multivariate count time series models are an important tool for the analysis
and prediction of infectious disease spread. We consider the endemic-epidemic
framework, an autoregressive model class for infectious disease surveillance
counts, and replace the default autoregression on counts from the previous time
period with more flexible weighting schemes inspired by discrete-time serial
interval distributions. We employ three different parametric formulations, each
with an additional unknown weighting parameter estimated via a profile
likelihood approach, and compare them to an unrestricted nonparametric
approach. The new methods are illustrated in a univariate analysis of dengue
fever incidence in San Juan, Puerto Rico, and a spatio-temporal study of viral
gastroenteritis in the twelve districts of Berlin. We assess the predictive
performance of the suggested models and several reference models at various
forecast horizons. In both applications, the performance of the
endemic-epidemic models is considerably improved by the proposed weighting
schemes.
| stat |
WAIC, but Why? Generative Ensembles for Robust Anomaly Detection | Machine learning models encounter Out-of-Distribution (OoD) errors when the
data seen at test time are generated from a different stochastic generator than
the one used to generate the training data. One proposal to scale OoD detection
to high-dimensional data is to learn a tractable likelihood approximation of
the training distribution, and use it to reject unlikely inputs. However,
likelihood models on natural data are themselves susceptible to OoD errors, and
even assign large likelihoods to samples from other datasets. To mitigate this
problem, we propose Generative Ensembles, which robustify density-based OoD
detection by way of estimating epistemic uncertainty of the likelihood model.
We present a puzzling observation in need of an explanation -- although
likelihood measures cannot account for the typical set of a distribution, and
therefore should not be suitable on their own for OoD detection, WAIC performs
surprisingly well in practice.
| stat |
Deriving Bounds and Inequality Constraints Using LogicalRelations Among
Counterfactuals | Causal parameters may not be point identified in the presence of unobserved
confounding. However, information about non-identified parameters, in the form
of bounds, may still be recovered from the observed data in some cases. We
develop a new general method for obtaining bounds on causal parameters using
rules of probability and restrictions on counterfactuals implied by causal
graphical models. We additionally provide inequality constraints on functionals
of the observed data law implied by such causal models. Our approach is
motivated by the observation that logical relations between identified and
non-identified counterfactual events often yield information about
non-identified events. We show that this approach is powerful enough to recover
known sharp bounds and tight inequality constraints, and to derive novel bounds
and constraints.
| stat |
Application of Cox Model to predict the survival of patients with
Chronic Heart Failure: A latent class regression approach | Most prediction models that are used in medical research fail to accurately
predict health outcomes due to methodological limitations. Using routinely
collected patient data, we explore the use of a Cox proportional hazard (PH)
model within a latent class framework to model survival of patients with
chronic heart failure (CHF). We identify subgroups of patients based on their
risk with the aid of available covariates. We allow each subgroup to have its
own risk model.We choose an optimum number of classes based on the reported
Bayesian information criteria (BIC). We assess the discriminative ability of
the chosen model using an area under the receiver operating characteristic
curve (AUC) for all the cross-validated and bootstrapped samples.We conduct a
simulation study to compare the predictive performance of our models. Our
proposed latent class model outperforms the standard one class Cox PH model.
| stat |
A Gaussian Process perspective on Convolutional Neural Networks | In this paper we cast the well-known convolutional neural network in a
Gaussian process perspective. In this way we hope to gain additional insights
into the performance of convolutional networks, in particular understand under
what circumstances they tend to perform well and what assumptions are
implicitly made in the network. While for fully-connected networks the
properties of convergence to Gaussian processes have been studied extensively,
little is known about situations in which the output from a convolutional
network approaches a multivariate normal distribution.
| stat |
Empirical Study of the Benefits of Overparameterization in Learning
Latent Variable Models | One of the most surprising and exciting discoveries in supervised learning
was the benefit of overparameterization (i.e. training a very large model) to
improving the optimization landscape of a problem, with minimal effect on
statistical performance (i.e. generalization). In contrast, unsupervised
settings have been under-explored, despite the fact that it was observed that
overparameterization can be helpful as early as Dasgupta & Schulman (2007). We
perform an empirical study of different aspects of overparameterization in
unsupervised learning of latent variable models via synthetic and
semi-synthetic experiments. We discuss benefits to different metrics of success
(recovering the parameters of the ground-truth model, held-out log-likelihood),
sensitivity to variations of the training algorithm, and behavior as the amount
of overparameterization increases. We find that across a variety of models
(noisy-OR networks, sparse coding, probabilistic context-free grammars) and
training algorithms (variational inference, alternating minimization,
expectation-maximization), overparameterization can significantly increase the
number of ground truth latent variables recovered.
| stat |
Generalizations of Edge Overlap to Weighted and Directed Networks | With the increasing availability of behavioral data from diverse digital
sources, such as social media sites and cell phones, it is now possible to
obtain detailed information about the structure, strength, and directionality
of social interactions in varied settings. While most metrics of network
structure have traditionally been defined for unweighted and undirected
networks only, the richness of current network data calls for extending these
metrics to weighted and directed networks. One fundamental metric in social
networks is edge overlap, the proportion of friends shared by two connected
individuals. Here we extend definitions of edge overlap to weighted and
directed networks, and present closed-form expressions for the mean and
variance of each version for the Erdos-Renyi random graph and its weighted and
directed counterparts. We apply these results to social network data collected
in rural villages in southern Karnataka, India. We use our analytical results
to quantify the extent to which the average overlap of the empirical social
network deviates from that of corresponding random graphs and compare the
values of overlap across networks. Our novel definitions allow the calculation
of edge overlap for more complex networks and our derivations provide a
statistically rigorous way for comparing edge overlap across networks.
| stat |
Inverse modeling of hydrologic parameters in CLM4 via generalized
polynomial chaos in the Bayesian framework | In this study, the applicability of generalized polynomial chaos (gPC)
expansion for land surface model parameter estimation is evaluated. We compute
the (posterior) distribution of the critical hydrological parameters that are
subject to great uncertainty in the community land model (CLM). The unknown
parameters include those that have been identified as the most influential
factors on the simulations of surface and subsurface runoff, latent and
sensible heat fluxes, and soil moisture in CLM4.0. We setup the inversion
problem this problem in the Bayesian framework in two steps: (i) build a
surrogate model expressing the input-output mapping, and (ii) compute the
posterior distributions of the input parameters. Development of the surrogate
model is done with a Bayesian procedure, based on the variable selection
methods that use gPC expansions. Our approach accounts for bases selection
uncertainty and quantifies the importance of the gPC terms, and hence all the
input parameters, via the associated posterior probabilities.
| stat |
Causal statistical modeling and calculation of distribution functions of
classification features | Statistical system models provide the basis for the examination of various
sorts of distributions. Classification distributions are a very common and
versatile form of statistics in e.g. real economic, social, and IT systems. The
statistical distributions of classification features can be applied in
determining the a priori probabilities in Bayesian networks. We investigate a
statistical model of classification distributions based on finding the critical
point of a specialized form of entropy. A distribution function for
classification features is derived, with the two parameters $n_0$, minimal
class, and $\bar{N}$, average number of classes. Efficient algorithms for the
computation of the class probabilities and the approximation of real frequency
distributions are developed and applied to examples from different domains. The
method is compared to established distributions like Zipf's law. The majority
of examples can be approximated with a sufficient quality ($3-5\%$).
| stat |
Bayesian Nonparametric Space Partitions: A Survey | Bayesian nonparametric space partition (BNSP) models provide a variety of
strategies for partitioning a $D$-dimensional space into a set of blocks. In
this way, the data points lie in the same block would share certain kinds of
homogeneity. BNSP models can be applied to various areas, such as
regression/classification trees, random feature construction, relational
modeling, etc. In this survey, we investigate the current progress of BNSP
research through the following three perspectives: models, which review various
strategies for generating the partitions in the space and discuss their
theoretical foundation `self-consistency'; applications, which cover the
current mainstream usages of BNSP models and their potential future practises;
and challenges, which identify the current unsolved problems and valuable
future research topics. As there are no comprehensive reviews of BNSP
literature before, we hope that this survey can induce further exploration and
exploitation on this topic.
| stat |
Feature Selection for Imbalanced Data with Deep Sparse Autoencoders
Ensemble | Class imbalance is a common issue in many domain applications of learning
algorithms. Oftentimes, in the same domains it is much more relevant to
correctly classify and profile minority class observations. This need can be
addressed by Feature Selection (FS), that offers several further advantages,
s.a. decreasing computational costs, aiding inference and interpretability.
However, traditional FS techniques may become sub-optimal in the presence of
strongly imbalanced data. To achieve FS advantages in this setting, we propose
a filtering FS algorithm ranking feature importance on the basis of the
Reconstruction Error of a Deep Sparse AutoEncoders Ensemble (DSAEE). We use
each DSAE trained only on majority class to reconstruct both classes. From the
analysis of the aggregated Reconstruction Error, we determine the features
where the minority class presents a different distribution of values w.r.t. the
overrepresented one, thus identifying the most relevant features to
discriminate between the two. We empirically demonstrate the efficacy of our
algorithm in several experiments on high-dimensional datasets of varying sample
size, showcasing its capability to select relevant and generalizable features
to profile and classify minority class, outperforming other benchmark FS
methods. We also briefly present a real application in radiogenomics, where the
methodology was applied successfully.
| stat |
BDNNSurv: Bayesian deep neural networks for survival analysis using
pseudo values | There has been increasing interest in modeling survival data using deep
learning methods in medical research. In this paper, we proposed a Bayesian
hierarchical deep neural networks model for modeling and prediction of survival
data. Compared with previously studied methods, the new proposal can provide
not only point estimate of survival probability but also quantification of the
corresponding uncertainty, which can be of crucial importance in predictive
modeling and subsequent decision making. The favorable statistical properties
of point and uncertainty estimates were demonstrated by simulation studies and
real data analysis. The Python code implementing the proposed approach was
provided.
| stat |
Classification of Categorical Time Series Using the Spectral Envelope
and Optimal Scalings | This article introduces a novel approach to the classification of categorical
time series under the supervised learning paradigm. To construct meaningful
features for categorical time series classification, we consider two relevant
quantities: the spectral envelope and its corresponding set of optimal
scalings. These quantities characterize oscillatory patterns in a categorical
time series as the largest possible power at each frequency, or spectral
envelope, obtained by assigning numerical values, or scalings, to categories
that optimally emphasize oscillations at each frequency. Our procedure combines
these two quantities to produce an interpretable and parsimonious feature-based
classifier that can be used to accurately determine group membership for
categorical time series. Classification consistency of the proposed method is
investigated, and simulation studies are used to demonstrate accuracy in
classifying categorical time series with various underlying group structures.
Finally, we use the proposed method to explore key differences in oscillatory
patterns of sleep stage time series for patients with different sleep disorders
and accurately classify patients accordingly.
| stat |
On the Optimality of Randomization in Experimental Design: How to
Randomize for Minimax Variance and Design-Based Inference | I study the minimax-optimal design for a two-arm controlled experiment where
conditional mean outcomes may vary in a given set. When this set is permutation
symmetric, the optimal design is complete randomization, and using a single
partition (i.e., the design that only randomizes the treatment labels for each
side of the partition) has minimax risk larger by a factor of $n-1$. More
generally, the optimal design is shown to be the mixed-strategy optimal design
(MSOD) of Kallus (2018). Notably, even when the set of conditional mean
outcomes has structure (i.e., is not permutation symmetric), being
minimax-optimal for variance still requires randomization beyond a single
partition. Nonetheless, since this targets precision, it may still not ensure
sufficient uniformity in randomization to enable randomization (i.e.,
design-based) inference by Fisher's exact test to appropriately detect
violations of null. I therefore propose the inference-constrained MSOD, which
is minimax-optimal among all designs subject to such uniformity constraints. On
the way, I discuss Johansson et al. (2020) who recently compared
rerandomization of Morgan and Rubin (2012) and the pure-strategy optimal design
(PSOD) of Kallus (2018). I point out some errors therein and set straight that
randomization is minimax-optimal and that the "no free lunch" theorem and
example in Kallus (2018) are correct.
| stat |
Optimal Balancing of Time-Dependent Confounders for Marginal Structural
Models | Marginal structural models (MSMs) estimate the causal effect of a
time-varying treatment in the presence of time-dependent confounding via
weighted regression. The standard approach of using inverse probability of
treatment weighting (IPTW) can lead to high-variance estimates due to extreme
weights and be sensitive to model misspecification. Various methods have been
proposed to partially address this, including truncation and stabilized-IPTW to
temper extreme weights and covariate balancing propensity score (CBPS) to
address treatment model misspecification. In this paper, we present Kernel
Optimal Weighting (KOW), a convex-optimization-based approach that finds
weights for fitting the MSM that optimally balance time-dependent confounders
while simultaneously controlling for precision, directly addressing the above
limitations. KOW directly minimizes the error in estimation due to
time-dependent confounding via a new decomposition as a functional. We further
extend KOW to control for informative censoring. We evaluate the performance of
KOW in a simulation study, comparing it with IPTW, stabilized-IPTW, and CBPS.
We demonstrate the use of KOW in studying the effect of treatment initiation on
time-to-death among people living with HIV and the effect of negative
advertising on elections in the United States.
| stat |
GO Gradient for Expectation-Based Objectives | Within many machine learning algorithms, a fundamental problem concerns
efficient calculation of an unbiased gradient wrt parameters $\gammav$ for
expectation-based objectives $\Ebb_{q_{\gammav} (\yv)} [f(\yv)]$. Most existing
methods either (i) suffer from high variance, seeking help from (often)
complicated variance-reduction techniques; or (ii) they only apply to
reparameterizable continuous random variables and employ a reparameterization
trick. To address these limitations, we propose a General and One-sample (GO)
gradient that (i) applies to many distributions associated with
non-reparameterizable continuous or discrete random variables, and (ii) has the
same low-variance as the reparameterization trick. We find that the GO gradient
often works well in practice based on only one Monte Carlo sample (although one
can of course use more samples if desired). Alongside the GO gradient, we
develop a means of propagating the chain rule through distributions, yielding
statistical back-propagation, coupling neural networks to common random
variables.
| stat |
Panel Data Quantile Regression for Treatment Effect Models | In this study, we develop a novel estimation method of the quantile treatment
effects (QTE) under the rank invariance and rank stationarity assumptions.
Ishihara (2020) explores identification of the nonseparable panel data model
under these assumptions and propose a parametric estimation based on the
minimum distance method. However, the minimum distance estimation using this
process is computationally demanding when the dimensionality of covariates is
large. To overcome this problem, we propose a two-step estimation method based
on the quantile regression and minimum distance method. We then show
consistency and asymptotic normality of our estimator. Monte Carlo studies
indicate that our estimator performs well in finite samples. Last, we present
two empirical illustrations, to estimate the distributional effects of
insurance provision on household production and of TV watching on child
cognitive development.
| stat |
Validation of Approximate Likelihood and Emulator Models for
Computationally Intensive Simulations | Complex phenomena in engineering and the sciences are often modeled with
computationally intensive feed-forward simulations for which a tractable
analytic likelihood does not exist. In these cases, it is sometimes necessary
to estimate an approximate likelihood or fit a fast emulator model for
efficient statistical inference; such surrogate models include Gaussian
synthetic likelihoods and more recently neural density estimators such as
autoregressive models and normalizing flows. To date, however, there is no
consistent way of quantifying the quality of such a fit. Here we propose a
statistical framework that can distinguish any arbitrary misspecified model
from the target likelihood, and that in addition can identify with statistical
confidence the regions of parameter as well as feature space where the fit is
inadequate. Our validation method applies to settings where simulations are
extremely costly and generated in batches or "ensembles" at fixed locations in
parameter space. At the heart of our approach is a two-sample test that
quantifies the quality of the fit at fixed parameter values, and a global test
that assesses goodness-of-fit across simulation parameters. While our general
framework can incorporate any test statistic or distance metric, we
specifically argue for a new two-sample test that can leverage any regression
method to attain high power and provide diagnostics in complex data settings.
| stat |
Adversarial robustness guarantees for random deep neural networks | The reliability of most deep learning algorithms is fundamentally challenged
by the existence of adversarial examples, which are incorrectly classified
inputs that are extremely close to a correctly classified input. We study
adversarial examples for deep neural networks with random weights and biases
and prove that the $\ell^1$ distance of any given input from the classification
boundary scales at least as $\sqrt{n}$, where $n$ is the dimension of the
input. We also extend our proof to cover all the $\ell^p$ norms. Our results
constitute a fundamental advance in the study of adversarial examples, and
encompass a wide variety of architectures, which include any combination of
convolutional or fully connected layers with skipped connections and pooling.
We validate our results with experiments on both random deep neural networks
and deep neural networks trained on the MNIST and CIFAR10 datasets. Given the
results of our experiments on MNIST and CIFAR10, we conjecture that the proof
of our adversarial robustness guarantee can be extended to trained deep neural
networks. This extension will open the way to a thorough theoretical study of
neural network robustness by classifying the relation between network
architecture and adversarial distance.
| stat |
A Novel Framework for Spatio-Temporal Prediction of Environmental Data
Using Deep Learning | As the role played by statistical and computational sciences in climate and
environmental modelling and prediction becomes more important, Machine Learning
researchers are becoming more aware of the relevance of their work to help
tackle the climate crisis. Indeed, being universal nonlinear function
approximation tools, Machine Learning algorithms are efficient in analysing and
modelling spatially and temporally variable environmental data. While Deep
Learning models have proved to be able to capture spatial, temporal, and
spatio-temporal dependencies through their automatic feature representation
learning, the problem of the interpolation of continuous spatio-temporal fields
measured on a set of irregular points in space is still under-investigated. To
fill this gap, we introduce here a framework for spatio-temporal prediction of
climate and environmental data using deep learning. Specifically, we show how
spatio-temporal processes can be decomposed in terms of a sum of products of
temporally referenced basis functions, and of stochastic spatial coefficients
which can be spatially modelled and mapped on a regular grid, allowing the
reconstruction of the complete spatio-temporal signal. Applications on two case
studies based on simulated and real-world data will show the effectiveness of
the proposed framework in modelling coherent spatio-temporal fields.
| stat |
Physics-Informed Gaussian Process Regression for Probabilistic States
Estimation and Forecasting in Power Grids | Real-time state estimation and forecasting is critical for efficient
operation of power grids. In this paper, a physics-informed Gaussian process
regression (PhI-GPR) method is presented and used for probabilistic forecasting
and estimating the phase angle, angular speed, and wind mechanical power of a
three-generator power grid system using sparse measurements. In standard
data-driven Gaussian process regression (GPR), parameterized models for the
prior statistics are fit by maximizing the marginal likelihood of observed
data, whereas in PhI-GPR, we compute the prior statistics by solving stochastic
equations governing power grid dynamics. The short-term forecast of a power
grid system dominated by wind generation is complicated by the stochastic
nature of the wind and the resulting uncertain mechanical wind power. Here, we
assume that the power-grid dynamic is governed by the swing equations, and we
treat the unknown terms in the swing equations (specifically, the mechanical
wind power) as random processes, which turns these equations into stochastic
differential equations. We solve these equations for the mean and variance of
the power grid system using the Monte Carlo simulations method. We demonstrate
that the proposed PhI-GPR method can accurately forecast and estimate both
observed and unobserved states, including the mean behavior and associated
uncertainty. For observed states, we show that PhI-GPR provides a forecast
comparable to the standard data-driven GPR, with both forecasts being
significantly more accurate than the autoregressive integrated moving average
(ARIMA) forecast. We also show that the ARIMA forecast is much more sensitive
to observation frequency and measurement errors than the PhI-GPR forecast.
| stat |
Stochastic Deep Networks | Machine learning is increasingly targeting areas where input data cannot be
accurately described by a single vector, but can be modeled instead using the
more flexible concept of random vectors, namely probability measures or more
simply point clouds of varying cardinality. Using deep architectures on
measures poses, however, many challenging issues. Indeed, deep architectures
are originally designed to handle fixedlength vectors, or, using recursive
mechanisms, ordered sequences thereof. In sharp contrast, measures describe a
varying number of weighted observations with no particular order. We propose in
this work a deep framework designed to handle crucial aspects of measures,
namely permutation invariances, variations in weights and cardinality.
Architectures derived from this pipeline can (i) map measures to measures -
using the concept of push-forward operators; (ii) bridge the gap between
measures and Euclidean spaces - through integration steps. This allows to
design discriminative networks (to classify or reduce the dimensionality of
input measures), generative architectures (to synthesize measures) and
recurrent pipelines (to predict measure dynamics). We provide a theoretical
analysis of these building blocks, review our architectures' approximation
abilities and robustness w.r.t. perturbation, and try them on various
discriminative and generative tasks.
| stat |
Testing multivariate normality by zeros of the harmonic oscillator in
characteristic function spaces | We study a novel class of affine invariant and consistent tests for normality
in any dimension. The tests are based on a characterization of the standard
$d$-variate normal distribution as the unique solution of an initial value
problem of a partial differential equation motivated by the harmonic
oscillator, which is a special case of a Schr\"odinger operator. We derive the
asymptotic distribution of the test statistics under the hypothesis of
normality as well as under fixed and contiguous alternatives. The tests are
consistent against general alternatives, exhibit strong power performance for
finite samples, and they are applied to a classical data set due to R.A.
Fisher. The results can also be used for a neighborhood-of-model validation
procedure.
| stat |
Analyzing the impact of two major factors on medical expenses paid by
health insurance organization in Iran | In healthcare scope, the profound role of insurance companies is undeniable.
Health insurance establishments main responsibility is to support public health
financially and promote the quality of health services. Governments subsidies
to healthcare insurance, insured payments and insurance companies costs must be
specified in such a way that both people and insurers mutually benefit. In this
research, we propose a model for determining healthcare costs paid by health
insurance organization with regard to two major factors, the geographical
regions where the patients live and the seasons when they receive service,
using two-way ANOVA method. Since both effects are found to be significant,
allocating different insurance costs to the people residing in different
regions, and also changing the patterns of insurance extension in different
seasons with regard to the results derived from the research, can detract
healthcare costs and give more satisfaction to the lower income insured
patients.
| stat |
Projection pursuit with applications to scRNA sequencing data | In this paper, we explore the limitations of PCA as a dimension reduction
technique and study its extension, projection pursuit (PP), which is a broad
class of linear dimension reduction methods. We first discuss the relevant
concepts and theorems and then apply PCA and PP (with negative standardized
Shannon's entropy as the projection index) on single cell RNA sequencing data.
| stat |
Unobserved classes and extra variables in high-dimensional discriminant
analysis | In supervised classification problems, the test set may contain data points
belonging to classes not observed in the learning phase. Moreover, the same
units in the test data may be measured on a set of additional variables
recorded at a subsequent stage with respect to when the learning sample was
collected. In this situation, the classifier built in the learning phase needs
to adapt to handle potential unknown classes and the extra dimensions. We
introduce a model-based discriminant approach, Dimension-Adaptive Mixture
Discriminant Analysis (D-AMDA), which can detect unobserved classes and adapt
to the increasing dimensionality. Model estimation is carried out via a full
inductive approach based on an EM algorithm. The method is then embedded in a
more general framework for adaptive variable selection and classification
suitable for data of large dimensions. A simulation study and an artificial
experiment related to classification of adulterated honey samples are used to
validate the ability of the proposed framework to deal with complex situations.
| stat |
CVXR: An R Package for Disciplined Convex Optimization | CVXR is an R package that provides an object-oriented modeling language for
convex optimization, similar to CVX, CVXPY, YALMIP, and Convex.jl. It allows
the user to formulate convex optimization problems in a natural mathematical
syntax rather than the restrictive form required by most solvers. The user
specifies an objective and set of constraints by combining constants,
variables, and parameters using a library of functions with known mathematical
properties. CVXR then applies signed disciplined convex programming (DCP) to
verify the problem's convexity. Once verified, the problem is converted into
standard conic form using graph implementations and passed to a cone solver
such as ECOS or SCS. We demonstrate CVXR's modeling framework with several
applications.
| stat |
Parsimonious Mixtures of Matrix Variate Bilinear Factor Analyzers | Over the years, data have become increasingly higher dimensional, which has
prompted an increased need for dimension reduction techniques. This is perhaps
especially true for clustering (unsupervised classification) as well as
semi-supervised and supervised classification. Many methods have been proposed
in the literature for two-way (multivariate) data and quite recently methods
have been presented for three-way (matrix variate) data. One such such method
is the mixtures of matrix variate bilinear factor analyzers (MMVBFA) model.
Herein, we propose of total of 64 parsimonious MMVBFA models. Simulated and
real data are used for illustration.
| stat |
Transmission of Macroeconomic Shocks to Risk Parameters: Their uses in
Stress Testing | In this paper, we are interested in evaluating the resilience of financial
portfolios under extreme economic conditions. Therefore, we use empirical
measures to characterize the transmission process of macroeconomic shocks to
risk parameters. We propose the use of an extensive family of models, called
General Transfer Function Models, which condense well the characteristics of
the transmission described by the impact measures. The procedure for estimating
the parameters of these models is described employing the Bayesian approach and
using the prior information provided by the impact measures. In addition, we
illustrate the use of the estimated models from the credit risk data of a
portfolio.
| stat |
New statistical control limits using maximum copula entropy | Statistical quality control methods are noteworthy to produced standard
production in manufacturing processes. In this regard, there are many classical
manners to control the process. Many of them have a global assumption around
distributions of the process data. They are supposed to be normal, which is
clear that it is not always valid for all processes. Such control charts made
some false decisions that waste funds. So, the main question while working with
multivariate data set is how to find the multivariate distribution of the data
set, which saves the original dependency between variables. Up to our
knowledge, a copula function guarantees the dependence on the result function.
But it is not enough when there is no other functional information about the
statistical society, and we have just a data set. Therefore, we apply the
maximum entropy concept to deal with this situation. In this paper, first of
all, we find out the joint distribution of a data set, which is from a
manufacturing process that needs to be control while running the production
process. Then, we get an elliptical control limit via the maximum copula
entropy. In the final step, we represent a practical example using the stated
method. Average run lengths are calculated for some means and shifts to show
the ability of the maximum copula entropy. In the end, two real data examples
are presented.
| stat |
Gibbs-type Indian buffet processes | We investigate a class of feature allocation models that generalize the
Indian buffet process and are parameterized by Gibbs-type random measures. Two
existing classes are contained as special cases: the original two-parameter
Indian buffet process, corresponding to the Dirichlet process, and the stable
(or three-parameter) Indian buffet process, corresponding to the Pitman--Yor
process. Asymptotic behavior of the Gibbs-type partitions, such as power laws
holding for the number of latent clusters, translates into analogous
characteristics for this class of Gibbs-type feature allocation models. Despite
containing several different distinct subclasses, the properties of Gibbs-type
partitions allow us to develop a black-box procedure for posterior inference
within any subclass of models. Through numerical experiments, we compare and
contrast a few of these subclasses and highlight the utility of varying
power-law behaviors in the latent features.
| stat |
A copula based approach for electoral quick counts | An electoral quick count is a statistical procedure whose main objective is
to obtain a relatively small but representative sample of all the polling
stations in a certain election, and to measure the uncertainty about the final
result before the total count of votes. A stratified sampling design is
commonly preferred to reduce estimation variability. The present work shows
that dependence among strata and among candidates should be taken into
consideration for statistical inferences therein, and a copula based model is
proposed and applied to Mexico's 2006, 2012, and 2018 presidential elections
data.
| stat |
External Correlates of Adult Digital Problem-Solving Behavior: Log Data
Analysis of a Large-Scale Assessment | Using the action sequence data (i.e., log data) from the problem-solving in
technology-rich environments assessment on the 2012 Programme for the
International Assessment of Adult Competencies survey, the current study
examines the associations between adult digital problem-solving behavior and
several demographic and cognitive variables. Action sequence features extracted
using multidimensional scaling (Tang, Wang, He, Liu, & Ying, 2019) and
sequence-to-sequence autoencoders (Tang, Wang, Liu, & Ying, 2019) were used to
predict test-taker external characteristics. Features extracted from action
sequences were consistently found to contain more information on demographic
and cognitive characteristics than final scores. Partial least squares analyses
further revealed systematic associations between behavioral patterns and
demographic/cognitive characteristics.
| stat |
A Regularized Vector Autoregressive Hidden Semi-Markov Model, with
Application to Multivariate Financial Data | A regularized vector autoregressive hidden semi-Markov model is developed to
analyze multivariate financial time series with switching data generating
regimes. Furthermore, an augmented EM algorithm is proposed for parameter
estimation by embedding regularized estimators for the state-dependent
covariance matrices and autoregression matrices in the M-step. The performance
of the proposed regularized estimators is evaluated both in the simulation
experiments and on the New York Stock Exchange financial portfolio data.
| stat |
An Online Updating Approach for Testing the Proportional Hazards
Assumption with Streams of Survival Data | The Cox model, which remains as the first choice in analyzing time-to-event
data even for large datasets, relies on the proportional hazards (PH)
assumption. When survival data arrive sequentially in chunks, a fast and
minimally storage intensive approach to test the PH assumption is desirable. We
propose an online updating approach that updates the standard test statistic as
each new block of data becomes available, and greatly lightens the
computational burden. Under the null hypothesis of PH, the proposed statistic
is shown to have the same asymptotic distribution as the standard version
computed on the entire data stream with the data blocks pooled into one
dataset. In simulation studies, the test and its variant based on most recent
data blocks maintain their sizes when the PH assumption holds and have
substantial power to detect different violations of the PH assumption. We also
show in simulation that our approach can be used successfully with "big data"
that exceed a single computer's computational resources. The approach is
illustrated with the survival analysis of patients with lymphoma cancer from
the Surveillance, Epidemiology, and End Results Program. The proposed test
promptly identified deviation from the PH assumption that was not captured by
the test based on the entire data.
| stat |
GGM knockoff filter: False Discovery Rate Control for Gaussian Graphical
Models | We propose a new method to learn the structure of a Gaussian graphical model
with finite sample false discovery rate control. Our method builds on the
knockoff framework of Barber and Cand\`{e}s for linear models. We extend their
approach to the graphical model setting by using a local (node-based) and a
global (graph-based) step: we construct knockoffs and feature statistics for
each node locally, and then solve a global optimization problem to determine a
threshold for each node. We then estimate the neighborhood of each node, by
comparing its feature statistics to its threshold, resulting in our graph
estimate. Our proposed method is very flexible, in the sense that there is
freedom in the choice of knockoffs, feature statistics, and the way in which
the final graph estimate is obtained. For any given data set, it is not clear a
priori what choices of these hyperparameters are optimal. We therefore use a
sample-splitting-recycling procedure that first uses half of the samples to
select the hyperparameters, and then learns the graph using all samples, in
such a way that the finite sample FDR control still holds. We compare our
method to several competitors in simulations and on a real data set.
| stat |
Adaptive Rate of Convergence of Thompson Sampling for Gaussian Process
Optimization | We consider the problem of global optimization of a function over a
continuous domain. In our setup, we can evaluate the function sequentially at
points of our choice and the evaluations are noisy. We frame it as a
continuum-armed bandit problem with a Gaussian Process prior on the function.
In this regime, most algorithms have been developed to minimize some form of
regret. In this paper, we study the convergence of the sequential point $x^t$
to the global optimizer $x^*$ for the Thompson Sampling approach. Under some
assumptions and regularity conditions, we prove concentration bounds for $x^t$
where the probability that $x^t$ is bounded away from $x^*$ decays
exponentially fast in $t$. Moreover, the result allows us to derive adaptive
convergence rates depending on the function structure.
| stat |
Rates of Convergence for Large-scale Nearest Neighbor Classification | Nearest neighbor is a popular class of classification methods with many
desirable properties. For a large data set which cannot be loaded into the
memory of a single machine due to computation, communication, privacy, or
ownership limitations, we consider the divide and conquer scheme: the entire
data set is divided into small subsamples, on which nearest neighbor
predictions are made, and then a final decision is reached by aggregating the
predictions on subsamples by majority voting. We name this method the big
Nearest Neighbor (bigNN) classifier, and provide its rates of convergence under
minimal assumptions, in terms of both the excess risk and the classification
instability, which are proven to be the same rates as the oracle nearest
neighbor classifier and cannot be improved. To significantly reduce the
prediction time that is required for achieving the optimal rate, we also
consider the pre-training acceleration technique applied to the bigNN method,
with proven convergence rate. We find that in the distributed setting, the
optimal choice of the neighbor $k$ should scale with both the total sample size
and the number of partitions, and there is a theoretical upper limit for the
latter. Numerical studies have verified the theoretical findings.
| stat |
Diagnostic tools for a multivariate negative binomial model for fitting
correlated data with overdispersion | We focus on the development of diagnostic tools and an R package called MNB
for a multivariate negative binomial (MNB) regression model for detecting
atypical and influential subjects. The MNB model is deduced from a Poisson
mixed model in which the random intercept follows the generalized log-gamma
(GLG) distribution. The MNB model for correlated count data leads to an MNB
regression model that inherits the features of a hierarchical model to
accommodate the intraclass correlation and the occurrence of overdispersion
simultaneously. The asymptotic consistency of the dispersion parameter
estimator depends on the asymmetry of the GLG distribution. Inferential
procedures for the MNB regression model are simple, although it can provide
inconsistent estimates of the asymptotic variance when the correlation
structure is misspecified. We propose the randomized quantile residual for
checking the adequacy of the multivariate model, and derive global and local
influence measures from the multivariate model to assess influential subjects.
Finally, two applications are presented in the data analysis section. The code
for installing the MNB package and the code used in the two examples is
exhibited in the Appendix.
| stat |
Outlier-Robust Spatial Perception: Hardness, General-Purpose Algorithms,
and Guarantees | Spatial perception is the backbone of many robotics applications, and spans a
broad range of research problems, including localization and mapping, point
cloud alignment, and relative pose estimation from camera images. Robust
spatial perception is jeopardized by the presence of incorrect data
association, and in general, outliers. Although techniques to handle outliers
do exist, they can fail in unpredictable manners (e.g., RANSAC, robust
estimators), or can have exponential runtime (e.g., branch-and-bound). In this
paper, we advance the state of the art in outlier rejection by making three
contributions. First, we show that even a simple linear instance of outlier
rejection is inapproximable: in the worst-case one cannot design a
quasi-polynomial time algorithm that computes an approximate solution
efficiently. Our second contribution is to provide the first per-instance
sub-optimality bounds to assess the approximation quality of a given outlier
rejection outcome. Our third contribution is to propose a simple
general-purpose algorithm, named adaptive trimming, to remove outliers. Our
algorithm leverages recently-proposed global solvers that are able to solve
outlier-free problems, and iteratively removes measurements with large errors.
We demonstrate the proposed algorithm on three spatial perception problems: 3D
registration, two-view geometry, and SLAM. The results show that our algorithm
outperforms several state-of-the-art methods across applications while being a
general-purpose method.
| stat |
Gaussian Process Latent Variable Alignment Learning | We present a model that can automatically learn alignments between
high-dimensional data in an unsupervised manner. Our proposed method casts
alignment learning in a framework where both alignment and data are modelled
simultaneously. Further, we automatically infer groupings of different types of
sequences within the same dataset. We derive a probabilistic model built on
non-parametric priors that allows for flexible warps while at the same time
providing means to specify interpretable constraints. We demonstrate the
efficacy of our approach with superior quantitative performance to the
state-of-the-art approaches and provide examples to illustrate the versatility
of our model in automatic inference of sequence groupings, absent from previous
approaches, as well as easy specification of high level priors for different
modalities of data.
| stat |
Distribution Matching in Variational Inference | With the increasingly widespread deployment of generative models, there is a
mounting need for a deeper understanding of their behaviors and limitations. In
this paper, we expose the limitations of Variational Autoencoders (VAEs), which
consistently fail to learn marginal distributions in both latent and visible
spaces. We show this to be a consequence of learning by matching conditional
distributions, and the limitations of explicit model and posterior
distributions. It is popular to consider Generative Adversarial Networks (GANs)
as a means of overcoming these limitations, leading to hybrids of VAEs and
GANs. We perform a large-scale evaluation of several VAE-GAN hybrids and
analyze the implications of class probability estimation for learning
distributions. While promising, we conclude that at present, VAE-GAN hybrids
have limited applicability: they are harder to scale, evaluate, and use for
inference compared to VAEs; and they do not improve over the generation quality
of GANs.
| stat |
Change-in-Slope Optimal Partitioning Algorithm in a Finite-Size
Parameter Space | We consider the problem of detecting change-points in univariate time series
by fitting a continuous piecewise linear signal using the residual sum of
squares. Values of the inferred signal at slope breaks are restricted to a
finite set of size $m$. Using this finite parameter space, we build a dynamic
programming algorithm with a controlled time complexity of $O(m^2n^2)$ for $n$
data points. Some accelerating strategies can be used to reduce the constant
before $n^2$. The adapted classic inequality-based pruning is outperformed by a
simpler "channel" method on simulations. Besides, our finite parameter space
setting allows an easy introduction of constraints on the inferred signal. For
example, imposing a minimal angle between consecutive segment slopes provides
robustness to model misspecification and outliers. We test our algorithm with
an isotonic constraint on an antibiogram image analysis problem for which
algorithmic efficiency is a cornerstone in the emerging context of
mobile-health and embedded medical devices. For this application, a finite
state approach can be a valid compromise.
| stat |
Unbiased Gradient Estimation for Distributionally Robust Learning | Seeking to improve model generalization, we consider a new approach based on
distributionally robust learning (DRL) that applies stochastic gradient descent
to the outer minimization problem. Our algorithm efficiently estimates the
gradient of the inner maximization problem through multi-level Monte Carlo
randomization. Leveraging theoretical results that shed light on why standard
gradient estimators fail, we establish the optimal parameterization of the
gradient estimators of our approach that balances a fundamental tradeoff
between computation time and statistical variance. Numerical experiments
demonstrate that our DRL approach yields significant benefits over previous
work.
| stat |
Binary disease prediction using tail quantiles of the distribution of
continuous biomarkers | In the analysis of binary disease classification, single biomarkers might not
have significant discriminating power and multiple biomarkers from a large set
of biomarkers should be selected. Numerous approaches exist, but they merely
work well for mean differences in biomarkers between cases and controls.
Biological processes are however much more heterogeneous, and differences could
also occur in other distributional characteristics (e.g. variances, skewness).
Many machine learning techniques are better capable of utilizing these higher
order distributional differences, sometimes at cost of explainability.
In this study we propose quantile based prediction (QBP), a binary
classification method that is based on the selection of multiple continuous
biomarkers. QBP generates a single score using the tails of the biomarker
distributions for cases and controls. This single score can then be evaluated
by ROC analysis to investigate its predictive power.
The performance of QBP is compared to supervised learning methods using
extensive simulation studies, and two case studies: major depression disorder
and trisomy. Simultaneously, the classification performance of the existing
techniques in relation to each other is assessed. The key strengths of QBP are
the opportunity to select relevant biomarkers and the outstanding
classification performance in the case biomarkers predominantly show variance
differences between cases and controls. When only shifts in means were present
in the biomarkers, QBP obtained an inferior performance. Lastly, QBP proved to
be unbiased in case of absence of disease relevant biomarkers and outperformed
the other methods on the MDD case study.
More research is needed to further optimize QBP, since it has several
opportunities to improve its performance. Here we wanted to introduce the
principle of QBP and show its potential.
| stat |
Obtaining interpretable parameters from reparameterizing longitudinal
models: transformation matrices between growth factors in two
parameter-spaces | The linear spline growth model (LSGM), which approximates complex patterns
using at least two linear segments, is a popular tool for examining nonlinear
change patterns. Among such models, the linear-linear piecewise change pattern
is the most straightforward one. An earlier study has proved that other than
the intercept and slopes, the knot (or change-point), at which two linear
segments join together, can be estimated as a growth factor in a
reparameterized longitudinal model in the latent growth curve modeling
framework. However, the reparameterized coefficients were no longer directly
related to the underlying developmental process and therefore lacked
meaningful, substantive interpretation, although they were simple functions of
the original parameters. This study proposes transformation matrices between
parameters in the original and reparameterized models so that the interpretable
coefficients directly related to the underlying change pattern can be derived
from reparameterized ones. Additionally, the study extends the existing
linear-linear piecewise model to allow for individual measurement occasions,
and investigates predictors for the individual-differences in change patterns.
We present the proposed methods with simulation studies and a real-world data
analysis. Our simulation studies demonstrate that the proposed method can
generally provide an unbiased and consistent estimation of model parameters of
interest and confidence intervals with satisfactory coverage probabilities. An
empirical example using longitudinal mathematics achievement scores shows that
the model can estimate the growth factor coefficients and path coefficients
directly related to the underlying developmental process, thereby providing
meaningful interpretation. For easier implementation, we also provide the
corresponding code for the proposed models.
| stat |
A Pseudo Knockoff Filter for Correlated Features | In 2015, Barber and Candes introduced a new variable selection procedure
called the knockoff filter to control the false discovery rate (FDR) and prove
that this method achieves exact FDR control. Inspired by the work of Barber and
Candes (2015), we propose and analyze a pseudo-knockoff filter that inherits
some advantages of the original knockoff filter and has more flexibility in
constructing its knockoff matrix. Moreover, we perform a number of numerical
experiments that seem to suggest that the pseudo knockoff filter with the half
Lasso statistic has FDR control and offers more power than the original
knockoff filter with the Lasso Path or the half Lasso Statistic for the
numerical examples that we consider in this paper. Although we cannot establish
rigorous FDR control for the pseudo knockoff filter, we provide some partial
analysis of the pseudo knockoff filter with the half Lasso statistic and
establish a uniform FDP bound and an expectation inequality.
| stat |
Linear regression with stationary errors : the R package slm | This paper introduces the R package slm which stands for Stationary Linear
Models. The package contains a set of statistical procedures for linear
regression in the general context where the error process is strictly
stationary with short memory. We work in the setting of Hannan (1973), who
proved the asymptotic normality of the (normalized) least squares estimators
(LSE) under very mild conditions on the error process. We propose different
ways to estimate the asymptotic covariance matrix of the LSE, and then to
correct the type I error rates of the usual tests on the parameters (as well as
confidence intervals). The procedures are evaluated through different sets of
simulations, and two examples of real datasets are studied.
| stat |
Spectrally-truncated kernel ridge regression and its free lunch | Kernel ridge regression (KRR) is a well-known and popular nonparametric
regression approach with many desirable properties, including minimax
rate-optimality in estimating functions that belong to common reproducing
kernel Hilbert spaces (RKHS). The approach, however, is computationally
intensive for large data sets, due to the need to operate on a dense $n \times
n$ kernel matrix, where $n$ is the sample size. Recently, various approximation
schemes for solving KRR have been considered, and some analyzed. Some
approaches such as Nystr\"{o}m approximation and sketching have been shown to
preserve the rate optimality of KRR. In this paper, we consider the simplest
approximation, namely, spectrally truncating the kernel matrix to its largest
$r < n$ eigenvalues. We derive an exact expression for the maximum risk of this
truncated KRR, over the unit ball of the RKHS. This result can be used to study
the exact trade-off between the level of spectral truncation and the
regularization parameter. We show that, as long as the RKHS is
infinite-dimensional, there is a threshold on $r$, above which, the
spectrally-truncated KRR surprisingly outperforms the full KRR in terms of the
minimax risk, where the minimum is taken over the regularization parameter.
This strengthens the existing results on approximation schemes, by showing that
not only one does not lose in terms of the rates, truncation can in fact
improve the performance, for all finite samples (above the threshold).
Moreover, we show that the implicit regularization achieved by spectral
truncation is not a substitute for Hilbert norm regularization. Both are needed
to achieve the best performance.
| stat |
Uncertainty Quantification in Multivariate Mixed Models for Mass
Cytometry Data | Mass cytometry technology enables the simultaneous measurement of over 40
proteins on single cells. This has helped immunologists to increase their
understanding of heterogeneity, complexity, and lineage relationships of white
blood cells. Current statistical methods often collapse the rich single-cell
data into summary statistics before proceeding with downstream analysis,
discarding the information in these multivariate datasets. In this article, our
aim is to exhibit the use of statistical analyses on the raw, uncompressed data
thus improving replicability, and exposing multivariate patterns and their
associated uncertainty profiles. We show that multivariate generative models
are a valid alternative to univariate hypothesis testing. We propose two
models: a multivariate Poisson log-normal mixed model and a logistic linear
mixed model. We show that these models are complementary and that either model
can account for different confounders. We use Hamiltonian Monte Carlo to
provide Bayesian uncertainty quantification. Our models applied to a recent
pregnancy study successfully reproduce key findings while quantifying increased
overall protein-to-protein correlations between first and third trimester.
| stat |
A Data-Driven Customer Segmentation Strategy Based on Contribution to
System Peak Demand | Advanced metering infrastructure (AMI) enables utilities to obtain granular
energy consumption data, which offers a unique opportunity to design customer
segmentation strategies based on their impact on various operational metrics in
distribution grids. However, performing utility-scale segmentation for
unobservable customers with only monthly billing information, remains a
challenging problem. To address this challenge, we propose a new metric, the
coincident monthly peak contribution (CMPC), that quantifies the contribution
of individual customers to system peak demand. Furthermore, a novel multi-state
machine learning-based segmentation method is developed that estimates CMPC for
customers without smart meters (SMs): first, a clustering technique is used to
build a databank containing typical daily load patterns in different seasons
using the SM data of observable customers. Next, to associate unobservable
customers with the discovered typical load profiles, a classification approach
is leveraged to compute the likelihood of daily consumption patterns for
different unobservable households. In the third stage, a weighted clusterwise
regression (WCR) model is utilized to estimate the CMPC of unobservable
customers using their monthly billing data and the outcomes of the
classification module. The proposed segmentation methodology has been tested
and verified using real utility data.
| stat |
Assessing the Frontier: Active Learning, Model Accuracy, and
Multi-objective Materials Discovery and Optimization | Discovering novel materials can be greatly accelerated by iterative machine
learning-informed proposal of candidates---active learning. However, standard
\emph{global-scope error} metrics for model quality are not predictive of
discovery performance, and can be misleading. We introduce the notion of
\emph{Pareto shell-scope error} to help judge the suitability of a model for
proposing material candidates. Further, through synthetic cases and a
thermoelectric dataset, we probe the relation between acquisition function
fidelity and active learning performance. Results suggest novel diagnostic
tools, as well as new insights for acquisition function design.
| stat |
Data Science in Biomedicine | We highlight the role of Data Science in Biomedicine. Our manuscript goes
from the general to the particular, presenting a global definition of Data
Science and showing the trend for this discipline together with the terms of
cloud computing and big data. In addition, since Data Science is mostly related
to areas like economy or business, we describe its importance in biomedicine.
Biomedical Data Science (BDS) presents the challenge of dealing with data
coming from a range of biological and medical research, focusing on
methodologies to advance the biomedical science discoveries, in an
interdisciplinary context.
| stat |
Home Sweet Home: Quantifying Home Court Advantages For NCAA Basketball
Statistics | Box score statistics are the baseline measures of performance for National
Collegiate Athletic Association (NCAA) basketball. Between the 2011-2012 and
2015-2016 seasons, NCAA teams performed better at home compared to on the road
in nearly all box score statistics across both genders and all three divisions.
Using box score data from over 100,000 games spanning the three divisions for
both women and men, we examine the factors underlying this discrepancy. The
prevalence of neutral location games in the NCAA provides an additional angle
through which to examine the gaps in box score statistic performance, which we
believe has been underutilized in existing literature. We also estimate a
regression model to quantify the home court advantages for box score statistics
after controlling for other factors such as number of possessions, and team
strength. Additionally, we examine the biases of scorekeepers and referees. We
present evidence that scorekeepers tend to have greater home team biases when
observing men compared to women, higher divisions compared to lower divisions,
and stronger teams compared to weaker teams. Finally, we present statistically
significant results indicating referee decisions are impacted by attendance,
with larger crowds resulting in greater bias in favor of the home team.
| stat |
Are official confirmed cases and fatalities counts good enough to study
the COVID-19 pandemic dynamics? A critical assessment through the case of
Italy | As the COVID-19 outbreak is developing the two most frequently reported
statistics seem to be the raw confirmed case and case fatalities counts.
Focusing on Italy, one of the hardest hit countries, we look at how these two
values could be put in perspective to reflect the dynamics of the virus spread.
In particular, we find that merely considering the confirmed case counts would
be very misleading. The number of daily tests grows, while the daily fraction
of confirmed cases to total tests has a change point. It (depending on region)
generally increases with strong fluctuations till (around, depending on region)
15th-22nd March and then decreases linearly after. Combined with the increasing
trend of daily performed tests, the raw confirmed case counts are not
representative of the situation and are confounded with the sampling effort.
This we observe when regressing on time the logged fraction of positive tests
and for comparison the logged raw confirmed count. Hence, calibrating model
parameters for this virus's dynamics should not be done based only on confirmed
case counts (without rescaling by the number of tests), but take also
fatalities and hospitalization count under consideration as variables not prone
to be distorted by testing efforts. Furthermore, reporting statistics on the
national level does not say much about the dynamics of the disease, which are
taking place at the regional level. These findings are based on the official
data of total death counts up to 15th April 2020 released by ISTAT and up to
10th May 2020 for the number of cases. In this work we do not fit models but we
rather investigate whether this task is possible at all. This work also informs
about a new tool to collect and harmonize official statistics coming from
different sources in the form of a package for the R statistical environment
and presents the COVID-19 Data Hub.
| stat |
Pandemic influenza and the gender imbalance: Evidence from early
twentieth century Japan | This study uses the 1918-1920 influenza pandemic in Japan with newly
digitized and complete census records on births, infant deaths, and sex ratios
during childhood to analyze mortality selection in utero and its persistency in
the gender imbalance. We find that fetal exposure to pandemic influenza during
the first trimester of the pregnancy period decreased the proportion of males
at birth in this period. We then show that the decline in male births might
have been associated with the deterioration of fetal health due to pandemic
influenza. Analyses using Population Censuses provide evidence suggesting that
postnatal influenza exposure had long-term impacts on the sex ratio of children
aged 5-12.
| stat |
Weekly Bayesian modelling strategy to predict deaths by COVID-19: a
model and case study for the state of Santa Catarina, Brazil | Background: The novel coronavirus pandemic has affected Brazil's Santa
Catarina State (SC) severely. At the time of writing (24 March 2021), over
764,000 cases and over 9,800 deaths by COVID-19 have been confirmed, hospitals
were fully occupied with local news reporting at least 397 people in the
waiting list for an ICU bed. In an attempt to better inform local policy
making, we applied an existing Bayesian algorithm to model the spread of the
pandemic in the seven geographic macro-regions of the state. Here we propose
changes to extend the model and improve its forecasting capabilities.
Methods: Our four proposed variations of the original method allow accessing
data of daily reported infections and take into account under-reporting of
cases more explicitly. Two of the proposed versions also attempt to model the
delay in test reporting. We simulated weekly forecasting of deaths from the
period from 31/05/2020 until 31/01/2021. First week data were used as a
cold-start to the algorithm, after which weekly calibrations of the model were
able to converge in fewer iterations. Google Mobility data were used as
covariates to the model, as well as to estimate of the susceptible population
at each simulated run.
Findings: The changes made the model significantly less reactive and more
rapid in adapting to scenarios after a peak in deaths is observed. Assuming
that the cases are under-reported greatly benefited the model in its stability,
and modelling retroactively-added data (due to the "hot" nature of the data
used) had a negligible impact in performance.
Interpretation: Although not as reliable as death statistics, case
statistics, when modelled in conjunction with an overestimate parameter,
provide a good alternative for improving the forecasting of models, especially
in long-range predictions and after the peak of an infection wave.
| stat |
Is there a cap on longevity? A statistical review | There is sustained and widespread interest in understanding the limit, if
any, to the human lifespan. Apart from its intrinsic and biological interest,
changes in survival in old age have implications for the sustainability of
social security systems. A central question is whether the endpoint of the
underlying lifetime distribution is finite. Recent analyses of data on the
oldest human lifetimes have led to competing claims about survival and to some
controversy, due in part to incorrect statistical analysis. This paper
discusses the particularities of such data, outlines correct ways of handling
them and presents suitable models and methods for their analysis. We provide a
critical assessment of some earlier work and illustrate the ideas through
reanalysis of semi-supercentenarian lifetime data. Our analysis suggests that
remaining life-length after age 109 is exponentially distributed, and that any
upper limit lies well beyond the highest lifetime yet reliably recorded. Lower
limits to 95% confidence intervals for the human lifespan are around 130 years,
and point estimates typically indicate no upper limit at all.
| stat |
Clustering Causal Additive Noise Models | Additive noise models are commonly used to infer the causal direction for a
given set of observed data. Most causal models assume a single homogeneous
population. However, observations may be collected under different conditions
in practice. Such data often require models that can accommodate possible
heterogeneity caused by different conditions under which data have been
collected. We propose a clustering algorithm inspired by the $k$-means
algorithm, but with unknown $k$. Using the proposed algorithm, both the labels
and the number of components are estimated from the collected data. The
estimated labels are used to adjust the causal direction test statistic. The
adjustment significantly improves the performance of the test statistic in
identifying the correct causal direction.
| stat |
Spatial Process Decomposition for Quantitative Imaging Biomarkers Using
Multiple Images of Varying Shapes | Quantitative imaging biomarkers (QIB) are extracted from medical images in
radiomics for a variety of purposes including noninvasive disease detection,
cancer monitoring, and precision medicine. The existing methods for QIB
extraction tend to be ad-hoc and not reproducible. In this paper, a general and
flexible statistical approach is proposed for handling up to three-dimensional
medical images in an objective and principled way. In particular, a model-based
spatial process decomposition is developed where the random weights are unique
to individual patients for component functions common across patients. Model
fitting and selection are based on maximum likelihood, while feature
extractions are via optimal prediction of the underlying true image. A
simulation study evaluates the properties of the proposed methodology and for
illustration, a cancer image data set is analyzed and QIBs are extracted in
association with a clinical endpoint.
| stat |
A Scale Invariant Flatness Measure for Deep Network Minima | It has been empirically observed that the flatness of minima obtained from
training deep networks seems to correlate with better generalization. However,
for deep networks with positively homogeneous activations, most measures of
sharpness/flatness are not invariant to rescaling of the network parameters,
corresponding to the same function. This means that the measure of
flatness/sharpness can be made as small or as large as possible through
rescaling, rendering the quantitative measures meaningless. In this paper we
show that for deep networks with positively homogenous activations, these
rescalings constitute equivalence relations, and that these equivalence
relations induce a quotient manifold structure in the parameter space. Using
this manifold structure and an appropriate metric, we propose a Hessian-based
measure for flatness that is invariant to rescaling. We use this new measure to
confirm the proposition that Large-Batch SGD minima are indeed sharper than
Small-Batch SGD minima.
| stat |
Bayesian beta nonlinear models with constrained parameters to describe
ruminal degradation kinetics | This paper proposes a beta nonlinear model to describe the kinetics of
ruminal degradation. The model generalizes the widely applied model proposed by
Orskov and McDonald (1979) because the proportion of degraded food is modelled
with the beta distribution and according to the Bayesian perspective. A default
method to obtain a prior distribution is proposed for this model, where the
application of standard methodologies, such as the Jeffreys prior (Jeffreys
(1961)) or the reference priors (Bernardo (1979)), involves serious
difficulties. This methodology is generalized to a larger class of models, and
an implementation of the method in OpenBUGS is shown.
| stat |
Active Structure Learning of Causal DAGs via Directed Clique Tree | A growing body of work has begun to study intervention design for efficient
structure learning of causal directed acyclic graphs (DAGs). A typical setting
is a causally sufficient setting, i.e. a system with no latent confounders,
selection bias, or feedback, when the essential graph of the observational
equivalence class (EC) is given as an input and interventions are assumed to be
noiseless. Most existing works focus on worst-case or average-case lower bounds
for the number of interventions required to orient a DAG. These worst-case
lower bounds only establish that the largest clique in the essential graph
could make it difficult to learn the true DAG. In this work, we develop a
universal lower bound for single-node interventions that establishes that the
largest clique is always a fundamental impediment to structure learning.
Specifically, we present a decomposition of a DAG into independently orientable
components through directed clique trees and use it to prove that the number of
single-node interventions necessary to orient any DAG in an EC is at least the
sum of half the size of the largest cliques in each chain component of the
essential graph. Moreover, we present a two-phase intervention design algorithm
that, under certain conditions on the chordal skeleton, matches the optimal
number of interventions up to a multiplicative logarithmic factor in the number
of maximal cliques. We show via synthetic experiments that our algorithm can
scale to much larger graphs than most of the related work and achieves better
worst-case performance than other scalable approaches. A code base to recreate
these results can be found at https://github.com/csquires/dct-policy
| stat |
A Comparative Study of Machine Learning Models for Predicting the State
of Reactive Mixing | Accurate predictions of reactive mixing are critical for many Earth and
environmental science problems. To investigate mixing dynamics over time under
different scenarios, a high-fidelity, finite-element-based numerical model is
built to solve the fast, irreversible bimolecular reaction-diffusion equations
to simulate a range of reactive-mixing scenarios. A total of 2,315 simulations
are performed using different sets of model input parameters comprising various
spatial scales of vortex structures in the velocity field, time-scales
associated with velocity oscillations, the perturbation parameter for the
vortex-based velocity, anisotropic dispersion contrast, and molecular
diffusion. Outputs comprise concentration profiles of the reactants and
products. The inputs and outputs of these simulations are concatenated into
feature and label matrices, respectively, to train 20 different machine
learning (ML) emulators to approximate system behavior. The 20 ML emulators
based on linear methods, Bayesian methods, ensemble learning methods, and
multilayer perceptron (MLP), are compared to assess these models. The ML
emulators are specifically trained to classify the state of mixing and predict
three quantities of interest (QoIs) characterizing species production, decay,
and degree of mixing. Linear classifiers and regressors fail to reproduce the
QoIs; however, ensemble methods (classifiers and regressors) and the MLP
accurately classify the state of reactive mixing and the QoIs. Among ensemble
methods, random forest and decision-tree-based AdaBoost faithfully predict the
QoIs. At run time, trained ML emulators are $\approx10^5$ times faster than the
high-fidelity numerical simulations. Speed and accuracy of the ensemble and MLP
models facilitate uncertainty quantification, which usually requires 1,000s of
model run, to estimate the uncertainty bounds on the QoIs.
| stat |
Position Dilution of Precision: a Bayesian point of view | The expected position error in many cases is far from feasible to be
estimated experimentally using real satellite measurements which makes the
model-based position dilution of precision (PDOP) crucial in positioning and
navigation applications. In the following text we derive the relationship
between PDOP and position error and we explain that this relationship holds as
long as the model for the observation errors represents the true sources of
errors.
| stat |
Automatic Model Building in GEFCom 2017 Qualifying Match | The Tangent Works team participated in GEFCom 2017 to test its automatic
model building strategy for time series known as Tangent Information Modeller
(TIM). Model building using TIM combined with historical temperature shuffling
resulted in winning the competition. This strategy involved one remaining
degree of freedom, a decision on using a trend variable. This paper describes
our modelling efforts in the competition, and furthermore outlines a fully
automated scenario where the decision on using the trend variable is handled by
TIM. The results show that such a setup would also win the competition.
| stat |
Entropy-SGD optimizes the prior of a PAC-Bayes bound: Generalization
properties of Entropy-SGD and data-dependent priors | We show that Entropy-SGD (Chaudhari et al., 2017), when viewed as a learning
algorithm, optimizes a PAC-Bayes bound on the risk of a Gibbs (posterior)
classifier, i.e., a randomized classifier obtained by a risk-sensitive
perturbation of the weights of a learned classifier. Entropy-SGD works by
optimizing the bound's prior, violating the hypothesis of the PAC-Bayes theorem
that the prior is chosen independently of the data. Indeed, available
implementations of Entropy-SGD rapidly obtain zero training error on random
labels and the same holds of the Gibbs posterior. In order to obtain a valid
generalization bound, we rely on a result showing that data-dependent priors
obtained by stochastic gradient Langevin dynamics (SGLD) yield valid PAC-Bayes
bounds provided the target distribution of SGLD is {\epsilon}-differentially
private. We observe that test error on MNIST and CIFAR10 falls within the
(empirically nonvacuous) risk bounds computed under the assumption that SGLD
reaches stationarity. In particular, Entropy-SGLD can be configured to yield
relatively tight generalization bounds and still fit real labels, although
these same settings do not obtain state-of-the-art performance.
| stat |
Advanced analysis of temporal data using Fisher-Shannon information:
theoretical development and application in geosciences | Complex non-linear time series are ubiquitous in geosciences. Quantifying
complexity and non-stationarity of these data is a challenging task, and
advanced complexity-based exploratory tool are required for understanding and
visualizing such data. This paper discusses the Fisher-Shannon method, from
which one can obtain a complexity measure and detect non-stationarity, as an
efficient data exploration tool. The state-of-the-art studies related to the
Fisher-Shannon measures are collected, and new analytical formulas for positive
unimodal skewed distributions are proposed. Case studies on both synthetic and
real data illustrate the usefulness of the Fisher-Shannon method, which can
find application in different domains including time series discrimination and
generation of times series features for clustering, modeling and forecasting.
The paper is accompanied with Python and R libraries for the non-parametric
estimation of the proposed measures.
| stat |
Outliers in meta-analysis: an asymmetric trimmed-mean approach | The adaptive asymmetric trimmed mean is a known way of estimating central
location, usually in conjunction with the bootstrap. It is here modified and
applied to meta-analysis, as a way of dealing with outlying results by
down-weighting the corresponding studies. This requires a modified bootstrap
and a method of down-weighting studies, as opposed to removing single
observations. This methodology is shown in analysis of some well-travelled
datasets to down-weight outliers in agreement with other methods, and
Monte-Carlo studies show that it does does not appreciably down-weight studies
when outliers are absent. Conceptually simple, it does not make parametric
assumptions about the outliers.
| stat |
Multifidelity Approximate Bayesian Computation | A vital stage in the mathematical modelling of real-world systems is to
calibrate a model's parameters to observed data. Likelihood-free parameter
inference methods, such as Approximate Bayesian Computation, build Monte Carlo
samples of the uncertain parameter distribution by comparing the data with
large numbers of model simulations. However, the computational expense of
generating these simulations forms a significant bottleneck in the practical
application of such methods. We identify how simulations of cheap, low-fidelity
models have been used separately in two complementary ways to reduce the
computational expense of building these samples, at the cost of introducing
additional variance to the resulting parameter estimates. We explore how these
approaches can be unified so that cost and benefit are optimally balanced, and
we characterise the optimal choice of how often to simulate from cheap,
low-fidelity models in place of expensive, high-fidelity models in Monte Carlo
ABC algorithms. The resulting early accept/reject multifidelity ABC algorithm
that we propose is shown to give improved performance over existing
multifidelity and high-fidelity approaches.
| stat |
A new estimator for Weibull distribution parameters: Comprehensive
comparative study for Weibull Distribution | Weibull distribution has received a wide range of applications in engineering
and science. The utility and usefulness of an estimator is highly subject to
the field of practitioner's study. In practice users looking for their desired
estimator under different setting of parameters and sample sizes. In this paper
we focus on two topics. Firstly, we propose $U$-statistics for the Weibull
distribution parameters. The consistency and asymptotically normality of the
introduced $U$-statistics are proved theoretically and by simulations. Several
of methods have been proposed for estimating the parameters of Weibull
distribution in the literature. These methods include: the generalized least
square type 1, the generalized least square type 2, the $L$-moments, the
Logarithmic moments, the maximum likelihood estimation, the method of moments,
the percentile method, the weighted least square, and weighted maximum
likelihood estimation. Secondary, due to lack of a comprehensive comparison
between the Weibull distribution parameters estimators, a comprehensive
comparison study is made between our proposed $U$-statistics and above nine
estimators. Based on simulations, it turns out that the our proposed
$U$-statistics show the best performance in terms of bias for estimating the
shape and scale parameters when the sample size is large.
| stat |
Complexity Results for MCMC derived from Quantitative Bounds | This paper considers how to obtain MCMC quantitative convergence bounds which
can be translated into tight complexity bounds in high-dimensional settings. We
propose a modified drift-and-minorization approach, which establishes a
generalized drift condition defined in a subset of the state space. The subset
is called the ``large set'' and is chosen to rule out some ``bad'' states which
have poor drift property when the dimension gets large. Using the ``large set''
together with a ``centered'' drift function, a quantitative bound can be
obtained which can be translated into a tight complexity bound. As a
demonstration, we analyze a certain realistic Gibbs sampler algorithm and
obtain a complexity upper bound for the mixing time, which shows that the
number of iterations required for the Gibbs sampler to converge is constant
under certain conditions on the observed data and the initial state. It is our
hope that this modified drift-and-minorization approach can be employed in many
other specific examples to obtain complexity bounds for high-dimensional Markov
chains.
| stat |
A New Class of Time Dependent Latent Factor Models with Applications | In many applications, observed data are influenced by some combination of
latent causes. For example, suppose sensors are placed inside a building to
record responses such as temperature, humidity, power consumption and noise
levels. These random, observed responses are typically affected by many
unobserved, latent factors (or features) within the building such as the number
of individuals, the turning on and off of electrical devices, power surges,
etc. These latent factors are usually present for a contiguous period of time
before disappearing; further, multiple factors could be present at a time. This
paper develops new probabilistic methodology and inference methods for random
object generation influenced by latent features exhibiting temporal
persistence. Every datum is associated with subsets of a potentially infinite
number of hidden, persistent features that account for temporal dynamics in an
observation. The ensuing class of dynamic models constructed by adapting the
Indian Buffet Process --- a probability measure on the space of random,
unbounded binary matrices --- finds use in a variety of applications arising in
operations, signal processing, biomedicine, marketing, image analysis, etc.
Illustrations using synthetic and real data are provided.
| stat |
New estimates for network sampling | Network sampling is used around the world for surveys of vulnerable,
hard-to-reach populations including people at risk for HIV, opioid misuse, and
emerging epidemics. The sampling methods include tracing social links to add
new people to the sample. Current estimates from these surveys are inaccurate,
with large biases and mean squared errors and unreliable confidence intervals.
New estimators are introduced here which eliminate almost all of the bias, have
much lower mean squared error, and enable confidence intervals with good
properties. The improvement is attained by avoiding unrealistic assumptions
about the population network and the design, instead using the topology of the
sample network data together with the sampling design actually used. In
simulations using the real network of an at-risk population, the new estimates
eliminate almost all the bias and have mean squared-errors that are 2 to 92
times lower than those of current estimators. The new estimators are effective
with a wide variety of network designs including those with strongly restricted
branching such as Respondent-Driven Sampling and freely branching designs such
as Snowball Sampling.
| stat |
Bayesian calibration of a numerical code for prediction | Field experiments are often difficult and expensive to make. To bypass these
issues, industrial companies have developed computational codes. These codes
intend to be representative of the physical system, but come with a certain
amount of problems. The code intends to be as close as possible to the physical
system. It turns out that, despite continuous code development, the difference
between the code outputs and experiments can remain significant. Two kinds of
uncertainties are observed. The first one comes from the difference between the
physical phenomenon and the values recorded experimentally. The second concerns
the gap between the code and the physical system. To reduce this difference,
often named model bias, discrepancy, or model error, computer codes are
generally complexified in order to make them more realistic. These improvements
lead to time consuming codes. Moreover, a code often depends on parameters to
be set by the user to make the code as close as possible to field data. This
estimation task is called calibration. This paper proposes a review of Bayesian
calibration methods and is based on an application case which makes it possible
to discuss the various methodological choices and to illustrate their
divergences. This example is based on a code used to predict the power of a
photovoltaic plant.
| stat |
Randomness is valid at large numbers | Randomness is a central concept to statistics and physics. Here, a
statistical analysis shows experimental evidence that tossing coins and finding
last digits of prime numbers are identical regarding statistics for equally
likely outcomes. This analysis explains why randomness in equally likely
outcomes can be valid only at large numbers.
| stat |
Ordering properties of the smallest and largest lifetimes in
Gompertz-Makeham model | The Gompertz-Makeham distribution, which is used commonly to represent
lifetimes based on laws of mortality, is one of the most popular choices for
mortality modelling in the field of actuarial science. This paper investigates
ordering properties of the smallest and largest lifetimes arising from two sets
of heterogeneous groups of insurees following respective Gompertz-Makeham
distributions. Some sufficient conditions are provided in the sense of usual
stochastic ordering to compare the smallest and largest lifetimes from two sets
of dependent variables. Comparison results on the smallest lifetimes in the
sense of hazard rate ordering and ageing faster ordering are established for
two groups of heterogeneous independent lifetimes. Under similar set-up, no
reversed hazard rate ordering is shown to exist between the largest lifetimes
with the use of a counter-example. Finally, we present sufficient conditions to
stochastically compare two sets of independent heterogeneous lifetimes under
random shocks by means of usual stochastic ordering. Such comparisons for the
smallest lifetimes are also carried out in terms of hazard rate ordering.
| stat |
An Empirical Evaluation of the Approximation of Subjective Logic
Operators Using Monte Carlo Simulations | In this paper we analyze the use of subjective logic as a framework for
performing approximate transformations over probability distribution functions.
As for any approximation, we evaluate subjective logic in terms of
computational efficiency and bias. However, while the computational cost may be
easily estimated, the bias of subjective logic operators have not yet been
investigated. In order to evaluate this bias, we propose an experimental
protocol that exploits Monte Carlo simulations and their properties to assess
the distance between the result produced by subjective logic operators and the
true result of the corresponding transformation over probability distributions.
This protocol allows a modeler to get an estimate of the degree of
approximation she must be ready to accept as a trade-off for the computational
efficiency and the interpretability of the subjective logic framework.
Concretely, we apply our method to the relevant case study of the subjective
logic operator for binomial multiplication and fusion, and we study empirically
their degree of approximation.
| stat |
Template Independent Component Analysis: Targeted and Reliable
Estimation of Subject-level Brain Networks using Big Data Population Priors | Large brain imaging databases contain a wealth of information on brain
organization in the populations they target, and on individual variability.
While such databases have been used to study group-level features of
populations directly, they are currently underutilized as a resource to inform
single-subject analysis. Here, we propose leveraging the information contained
in large functional magnetic resonance imaging (fMRI) databases by establishing
population priors to employ in an empirical Bayesian framework. We focus on
estimation of brain networks as source signals in independent component
analysis (ICA). We formulate a hierarchical "template" ICA model where source
signals---including known population brain networks and subject-specific
signals---are represented as latent variables. For estimation, we derive an
expectation maximization (EM) algorithm having an explicit solution. However,
as this solution is computationally intractable, we also consider an
approximate subspace algorithm and a faster two-stage approach. Through
extensive simulation studies, we assess performance of both methods and compare
with dual regression, a popular but ad-hoc method. The two proposed algorithms
have similar performance, and both dramatically outperform dual regression. We
also conduct a reliability study utilizing the Human Connectome Project and
find that template ICA achieves substantially better performance than dual
regression, achieving 75-250% higher intra-subject reliability.
| stat |
Information Theory Measures via Multidimensional Gaussianization | Information theory is an outstanding framework to measure uncertainty,
dependence and relevance in data and systems. It has several desirable
properties for real world applications: it naturally deals with multivariate
data, it can handle heterogeneous data types, and the measures can be
interpreted in physical units. However, it has not been adopted by a wider
audience because obtaining information from multidimensional data is a
challenging problem due to the curse of dimensionality. Here we propose an
indirect way of computing information based on a multivariate Gaussianization
transform. Our proposal mitigates the difficulty of multivariate density
estimation by reducing it to a composition of tractable (marginal) operations
and simple linear transformations, which can be interpreted as a particular
deep neural network. We introduce specific Gaussianization-based methodologies
to estimate total correlation, entropy, mutual information and Kullback-Leibler
divergence. We compare them to recent estimators showing the accuracy on
synthetic data generated from different multivariate distributions. We made the
tools and datasets publicly available to provide a test-bed to analyze future
methodologies. Results show that our proposal is superior to previous
estimators particularly in high-dimensional scenarios; and that it leads to
interesting insights in neuroscience, geoscience, computer vision, and machine
learning.
| stat |
General Regression Methods for Respondent-Driven Sampling Data | Respondent-Driven Sampling (RDS) is a variant of link-tracing sampling
techniques that aim to recruit hard-to-reach populations by leveraging
individuals' social relationships. As such, an RDS sample has a graphical
component which represents a partially observed network of unknown structure.
Moreover, it is common to observe homophily, or the tendency to form
connections with individuals who share similar traits. Currently, there is a
lack of principled guidance on multivariate modeling strategies for RDS to
address homophilic covariates and the dependence between observations within
the network. In this work, we propose a methodology for general regression
techniques using RDS data. This is used to study the socio-demographic
predictors of HIV treatment optimism (about the value of antiretroviral
therapy) among gay, bisexual and other men who have sex with men, recruited
into an RDS study in Montreal, Canada.
| stat |
The Efficient Shrinkage Path: Maximum Likelihood of Minimum MSE Risk | A new generalized ridge regression shrinkage path is proposed that is as
short as possible under the restriction that it must pass through the vector of
regression coefficient estimators that make the overall Optimal Variance-Bias
Trade-Off under Normal distribution-theory. Five distinct types of ridge TRACE
displays and other graphics for this efficient path are motivated and
illustrated here. These visualizations provide invaluable data-analytic
insights and improved self-confidence to researchers and data scientists
fitting linear models to ill-conditioned (confounded) data.
| stat |
Large Non-Stationary Noisy Covariance Matrices: A Cross-Validation
Approach | We introduce a novel covariance estimator that exploits the heteroscedastic
nature of financial time series by employing exponential weighted moving
averages and shrinking the in-sample eigenvalues through cross-validation. Our
estimator is model-agnostic in that we make no assumptions on the distribution
of the random entries of the matrix or structure of the covariance matrix.
Additionally, we show how Random Matrix Theory can provide guidance for
automatic tuning of the hyperparameter which characterizes the time scale for
the dynamics of the estimator. By attenuating the noise from both the
cross-sectional and time-series dimensions, we empirically demonstrate the
superiority of our estimator over competing estimators that are based on
exponentially-weighted and uniformly-weighted covariance matrices.
| stat |
Automatic Induction of Neural Network Decision Tree Algorithms | This work presents an approach to automatically induction for non-greedy
decision trees constructed from neural network architecture. This construction
can be used to transfer weights when growing or pruning a decision tree,
allowing non-greedy decision tree algorithms to automatically learn and adapt
to the ideal architecture. In this work, we examine the underpinning ideas
within ensemble modelling and Bayesian model averaging which allow our neural
network to asymptotically approach the ideal architecture through weights
transfer. Experimental results demonstrate that this approach improves models
over fixed set of hyperparameters for decision tree models and decision forest
models.
| stat |
Subsets and Splits
No saved queries yet
Save your SQL queries to embed, download, and access them later. Queries will appear here once saved.