title
stringlengths 6
244
| abstract
stringlengths 19
5.09k
| label
stringclasses 10
values |
|---|---|---|
Probabilistic programming for birth-death models of evolution using an
alive particle filter with delayed sampling | We consider probabilistic programming for birth-death models of evolution and
introduce a new widely-applicable inference method that combines an extension
of the alive particle filter (APF) with automatic Rao-Blackwellization via
delayed sampling. Birth-death models of evolution are an important family of
phylogenetic models of the diversification processes that lead to evolutionary
trees. Probabilistic programming languages (PPLs) give phylogeneticists a new
and exciting tool: their models can be implemented as probabilistic programs
with just a basic knowledge of programming. The general inference methods in
PPLs reduce the need for external experts, allow quick prototyping and testing,
and accelerate the development and deployment of new models. We show how these
birth-death models can be implemented as simple programs in existing PPLs, and
demonstrate the usefulness of the proposed inference method for such models.
For the popular BiSSE model the method yields an increase of the effective
sample size and the conditional acceptance rate by a factor of 30 in comparison
with a standard bootstrap particle filter. Although concentrating on
phylogenetics, the extended APF is a general inference method that shows its
strength in situations where particles are often assigned zero weight. In the
case when the weights are always positive, the extra cost of using the APF
rather than the bootstrap particle filter is negligible, making our method a
suitable drop-in replacement for the bootstrap particle filter in probabilistic
programming inference.
| stat |
Spatial Multivariate Trees for Big Data Bayesian Regression | High resolution geospatial data are challenging because standard
geostatistical models based on Gaussian processes are known to not scale to
large data sizes. While progress has been made towards methods that can be
computed more efficiently, considerably less attention has been devoted to big
data methods that allow the description of complex relationships between
several outcomes recorded at high resolutions by different sensors. Our
Bayesian multivariate regression models based on spatial multivariate trees
(SpamTrees) achieve scalability via conditional independence assumptions on
latent random effects following a treed directed acyclic graph.
Information-theoretic arguments and considerations on computational efficiency
guide the construction of the tree and the related efficient sampling
algorithms in imbalanced multivariate settings. In addition to simulated data
examples, we illustrate SpamTrees using a large climate data set which combines
satellite data with land-based station data. Source code is available at
https://github.com/mkln/spamtree
| stat |
Almost-Matching-Exactly for Treatment Effect Estimation under Network
Interference | We propose a matching method that recovers direct treatment effects from
randomized experiments where units are connected in an observed network, and
units that share edges can potentially influence each others' outcomes.
Traditional treatment effect estimators for randomized experiments are biased
and error prone in this setting. Our method matches units almost exactly on
counts of unique subgraphs within their neighborhood graphs. The matches that
we construct are interpretable and high-quality. Our method can be extended
easily to accommodate additional unit-level covariate information. We show
empirically that our method performs better than other existing methodologies
for this problem, while producing meaningful, interpretable results.
| stat |
Predicting Shot Making in Basketball Learnt from Adversarial Multiagent
Trajectories | In this paper, we predict the likelihood of a player making a shot in
basketball from multiagent trajectories. Previous approaches to similar
problems center on hand-crafting features to capture domain specific knowledge.
Although intuitive, recent work in deep learning has shown this approach is
prone to missing important predictive features. To circumvent this issue, we
present a convolutional neural network (CNN) approach where we initially
represent the multiagent behavior as an image. To encode the adversarial nature
of basketball, we use a multi-channel image which we then feed into a CNN.
Additionally, to capture the temporal aspect of the trajectories we "fade" the
player trajectories. We find that this approach is superior to a traditional
FFN model. By using gradient ascent to create images using an already trained
CNN, we discover what features the CNN filters learn. Last, we find that a
combined CNN+FFN is the best performing network with an error rate of 39%.
| stat |
Deductron -- A Recurrent Neural Network | The current paper is a study in Recurrent Neural Networks (RNN), motivated by
the lack of examples simple enough so that they can be thoroughly understood
theoretically, but complex enough to be realistic. We constructed an example of
structured data, motivated by problems from image-to-text conversion (OCR),
which requires long-term memory to decode. Our data is a simple writing system,
encoding characters 'X' and 'O' as their upper halves, which is possible due to
symmetry of the two characters. The characters can be connected, as in some
languages using cursive, such as Arabic (abjad). The string 'XOOXXO' may be
encoded as
'${\vee}{\wedge}\kern-1.5pt{\wedge}{\vee}\kern-1.5pt{\vee}{\wedge}$'. It
follows that we may need to know arbitrarily long past to decode a current
character, thus requiring long-term memory. Subsequently we constructed an RNN
capable of decoding sequences encoded in this manner. Rather than by training,
we constructed our RNN "by inspection", i.e. we guessed its weights. This
involved a sequence of steps. We wrote a conventional program which decodes the
sequences as the example above. Subsequently, we interpreted the program as a
neural network (the only example of this kind known to us). Finally, we
generalized this neural network to discover a new RNN architecture whose
instance is our handcrafted RNN. It turns out to be a 3 layer network, where
the middle layer is capable of performing simple logical inferences; thus the
name "deductron". It is demonstrated that it is possible to train our network
by simulated annealing. Also, known variants of stochastic gradient descent
(SGD) methods are shown to work.
| stat |
Subsidising Inclusive Insurance to Reduce Poverty | In this article, we consider a compound Poisson-type model for households'
capital. Using risk theory techniques, we determine the probability of a
household falling under the poverty line. Microinsurance is then introduced to
analyse its impact as an insurance solution for the lower income class. Our
results validate those previously obtained with this type of model, showing
that microinsurance alone is not sufficient to reduce the probability of
falling into the area of poverty for specific groups of people, since premium
payments constrain households' capital growth. This indicates the need for
additional aid particularly from the government. As such, we propose several
premium subsidy strategies and discuss the role of government in subsidising
microinsurance to help reduce poverty.
| stat |
Stability and Generalization of the Decentralized Stochastic Gradient
Descent | The stability and generalization of stochastic gradient-based methods provide
valuable insights into understanding the algorithmic performance of machine
learning models. As the main workhorse for deep learning, stochastic gradient
descent has received a considerable amount of studies. Nevertheless, the
community paid little attention to its decentralized variants. In this paper,
we provide a novel formulation of the decentralized stochastic gradient
descent. Leveraging this formulation together with (non)convex optimization
theory, we establish the first stability and generalization guarantees for the
decentralized stochastic gradient descent. Our theoretical results are built on
top of a few common and mild assumptions and reveal that the decentralization
deteriorates the stability of SGD for the first time. We verify our theoretical
findings by using a variety of decentralized settings and benchmark machine
learning models.
| stat |
Estimating the Variance of Measurement Errors in Running Variables of
Sharp Regression Discontinuity Designs | Estimation of a treatment effect by a regression discontinuity design faces a
severe challenge when the running variable contains measurement errors since
the errors smoothen the discontinuity on which the identification depends. The
existing studies show that the variance of the measurement errors plays a vital
role in both bias correction and identification under such situations. However,
the methodologies to estimate the variance from data are relatively
undeveloped. This paper proposes two estimators for the variance of measurement
errors of running variables of sharp regression continuity designs. The
proposed estimators can be constructed merely from data of observed running
variable and treatment assignment, and do not require any other external source
of information.
| stat |
Are Clusterings of Multiple Data Views Independent? | In the Pioneer 100 (P100) Wellness Project (Price and others, 2017), multiple
types of data are collected on a single set of healthy participants at multiple
timepoints in order to characterize and optimize wellness. One way to do this
is to identify clusters, or subgroups, among the participants, and then to
tailor personalized health recommendations to each subgroup. It is tempting to
cluster the participants using all of the data types and timepoints, in order
to fully exploit the available information. However, clustering the
participants based on multiple data views implicitly assumes that a single
underlying clustering of the participants is shared across all data views. If
this assumption does not hold, then clustering the participants using multiple
data views may lead to spurious results. In this paper, we seek to evaluate the
assumption that there is some underlying relationship among the clusterings
from the different data views, by asking the question: are the clusters within
each data view dependent or independent? We develop a new test for answering
this question, which we then apply to clinical, proteomic, and metabolomic
data, across two distinct timepoints, from the P100 study. We find that while
the subgroups of the participants defined with respect to any single data type
seem to be dependent across time, the clustering among the participants based
on one data type (e.g. proteomic data) appears not to be associated with the
clustering based on another data type (e.g. clinical data).
| stat |
Bayesian inference for transportation origin-destination matrices: the
Poisson-inverse Gaussian and other Poisson mixtures | In this paper we present Poisson mixture approaches for origin-destination
(OD) modeling in transportation analysis. We introduce covariate-based models
which incorporate different transport modeling phases and also allow for direct
probabilistic inference on link traffic based on Bayesian predictions. Emphasis
is placed on the Poisson-inverse Gaussian as an alternative to the
commonly-used Poisson-gamma and Poisson-lognormal models. We present a first
full Bayesian formulation and demonstrate that the Poisson-inverse Gaussian is
particularly suited for OD analysis due to desirable marginal and hierarchical
properties. In addition, the integrated nested Laplace approximation (INLA) is
considered as an alternative to Markov chain Monte Carlo and the two
methodologies are compared under specific modeling assumptions. The case study
is based on 2001 Belgian census data and focuses on a large,
sparsely-distributed OD matrix containing trip information for 308 Flemish
municipalities.
| stat |
A new Bayesian two-sample t-test for effect size estimation under
uncertainty based on a two-component Gaussian mixture with known allocations
and the region of practical equivalence | Testing differences between a treatment and control group is common practice
in biomedical research like randomized controlled trials (RCT). The standard
two-sample t-test relies on null hypothesis significance testing (NHST) via
p-values, which has several drawbacks. Bayesian alternatives were recently
introduced using the Bayes factor, which has its own limitations. This paper
introduces an alternative to current Bayesian two-sample t-tests by
interpreting the underlying model as a two-component Gaussian mixture in which
the effect size is the quantity of interest, which is most relevant in clinical
research. Unlike p-values or the Bayes factor, the proposed method focusses on
estimation under uncertainty instead of explicit hypothesis testing. Therefore,
via a Gibbs sampler the posterior of the effect size is produced, which is used
subsequently for either estimation under uncertainty or explicit hypothesis
testing based on the region of practical equivalence (ROPE). An illustrative
example, theoretical results and a simulation study show the usefulness of the
proposed method, and the test is made available in the R package bayest.
| stat |
Assessing and Visualizing Simultaneous Simulation Error | Monte Carlo experiments produce samples in order to estimate features of a
given distribution. However, simultaneous estimation of means and quantiles has
received little attention, despite being common practice. In this setting we
establish a multivariate central limit theorem for any finite combination of
sample means and quantiles under the assumption of a strongly mixing process,
which includes the standard Monte Carlo and Markov chain Monte Carlo settings.
We build on this to provide a fast algorithm for constructing hyperrectangular
confidence regions having the desired simultaneous coverage probability and a
convenient marginal interpretation. The methods are incorporated into standard
ways of visualizing the results of Monte Carlo experiments enabling the
practitioner to more easily assess the reliability of the results. We
demonstrate the utility of this approach in various Monte Carlo settings
including simulation studies based on independent and identically distributed
samples and Bayesian analyses using Markov chain Monte Carlo sampling.
| stat |
Variational Inference for Stochastic Block Models from Sampled Data | This paper deals with non-observed dyads during the sampling of a network and
consecutive issues in the inference of the Stochastic Block Model (SBM). We
review sampling designs and recover Missing At Random (MAR) and Not Missing At
Random (NMAR) conditions for the SBM. We introduce variants of the variational
EM algorithm for inferring the SBM under various sampling designs (MAR and
NMAR) all available as an R package. Model selection criteria based on
Integrated Classification Likelihood are derived for selecting both the number
of blocks and the sampling design. We investigate the accuracy and the range of
applicability of these algorithms with simulations. We explore two real-world
networks from ethnology (seed circulation network) and biology (protein-protein
interaction network), where the interpretations considerably depends on the
sampling designs considered.
| stat |
Deep Direct Likelihood Knockoffs | Predictive modeling often uses black box machine learning methods, such as
deep neural networks, to achieve state-of-the-art performance. In scientific
domains, the scientist often wishes to discover which features are actually
important for making the predictions. These discoveries may lead to costly
follow-up experiments and as such it is important that the error rate on
discoveries is not too high. Model-X knockoffs enable important features to be
discovered with control of the FDR. However, knockoffs require rich generative
models capable of accurately modeling the knockoff features while ensuring they
obey the so-called "swap" property. We develop Deep Direct Likelihood Knockoffs
(DDLK), which directly minimizes the KL divergence implied by the knockoff swap
property. DDLK consists of two stages: it first maximizes the explicit
likelihood of the features, then minimizes the KL divergence between the joint
distribution of features and knockoffs and any swap between them. To ensure
that the generated knockoffs are valid under any possible swap, DDLK uses the
Gumbel-Softmax trick to optimize the knockoff generator under the worst-case
swap. We find DDLK has higher power than baselines while controlling the false
discovery rate on a variety of synthetic and real benchmarks including a task
involving a large dataset from one of the epicenters of COVID-19.
| stat |
Stochastic filters based on hybrid approximations of multiscale
stochastic reaction networks | We consider the problem of estimating the dynamic latent states of an
intracellular multiscale stochastic reaction network from time-course
measurements of fluorescent reporters. We first prove that accurate solutions
to the filtering problem can be constructed by solving the filtering problem
for a reduced model that represents the dynamics as a hybrid process. The model
reduction is based on exploiting the time-scale separations in the original
network, and it can greatly reduce the computational effort required to
simulate the dynamics. This enables us to develop efficient particle filters to
solve the filtering problem for the original model by applying particle filters
to the reduced model. We illustrate the accuracy and the computational
efficiency of our approach using a numerical example.
| stat |
A Bayesian Redesign of the First Probability/Statistics Course | The traditional calculus-based introduction to statistical inference consists
of a semester of probability followed by a semester of frequentist inference.
Cobb (2015) challenges the statistical education community to rethink the
undergraduate statistics curriculum. In particular, he suggests that we should
focus on two goals: making fundamental concepts accessible and minimizing
prerequisites to research. Using five underlying principles of Cobb, we
describe a new calculus-based introduction to statistics based on
simulation-based Bayesian computation.
| stat |
Stable Rank Normalization for Improved Generalization in Neural Networks
and GANs | Exciting new work on the generalization bounds for neural networks (NN) given
by Neyshabur et al. , Bartlett et al. closely depend on two
parameter-depenedent quantities: the Lipschitz constant upper-bound and the
stable rank (a softer version of the rank operator). This leads to an
interesting question of whether controlling these quantities might improve the
generalization behaviour of NNs. To this end, we propose stable rank
normalization (SRN), a novel, optimal, and computationally efficient
weight-normalization scheme which minimizes the stable rank of a linear
operator. Surprisingly we find that SRN, inspite of being non-convex problem,
can be shown to have a unique optimal solution. Moreover, we show that SRN
allows control of the data-dependent empirical Lipschitz constant, which in
contrast to the Lipschitz upper-bound, reflects the true behaviour of a model
on a given dataset. We provide thorough analyses to show that SRN, when applied
to the linear layers of a NN for classification, provides striking
improvements-11.3% on the generalization gap compared to the standard NN along
with significant reduction in memorization. When applied to the discriminator
of GANs (called SRN-GAN) it improves Inception, FID, and Neural divergence
scores on the CIFAR 10/100 and CelebA datasets, while learning mappings with
low empirical Lipschitz constants.
| stat |
Co-clustering of time-dependent data via Shape Invariant Model | Multivariate time-dependent data, where multiple features are observed over
time for a set of individuals, are increasingly widespread in many application
domains. To model these data we need to account for relations among both time
instants and variables and, at the same time, for subjects heterogeneity. We
propose a new co-clustering methodology for clustering individuals and
variables simultaneously that is designed to handle both functional and
longitudinal data. Our approach borrows some concepts from the curve
registration framework by embedding the Shape Invariant Model in the Latent
Block Model, estimated via a suitable modification of the SEM-Gibbs algorithm.
The resulting procedure allows for several user-defined specifications of the
notion of cluster that could be chosen on substantive grounds and provides
parsimonious summaries of complex longitudinal or functional data by
partitioning data matrices into homogeneous blocks.
| stat |
Metrics for Graph Comparison: A Practitioner's Guide | Comparison of graph structure is a ubiquitous task in data analysis and
machine learning, with diverse applications in fields such as neuroscience,
cyber security, social network analysis, and bioinformatics, among others.
Discovery and comparison of structures such as modular communities, rich clubs,
hubs, and trees in data in these fields yields insight into the generative
mechanisms and functional properties of the graph.
Often, two graphs are compared via a pairwise distance measure, with a small
distance indicating structural similarity and vice versa. Common choices
include spectral distances (also known as $\lambda$ distances) and distances
based on node affinities. However, there has of yet been no comparative study
of the efficacy of these distance measures in discerning between common graph
topologies and different structural scales.
In this work, we compare commonly used graph metrics and distance measures,
and demonstrate their ability to discern between common topological features
found in both random graph models and empirical datasets. We put forward a
multi-scale picture of graph structure, in which the effect of global and local
structure upon the distance measures is considered. We make recommendations on
the applicability of different distance measures to empirical graph data
problem based on this multi-scale view. Finally, we introduce the Python
library NetComp which implements the graph distances used in this work.
| stat |
Timber Volume Estimation Based on Airborne Laser Scanning -- Comparing
the Use of National Forest Inventory and Forest Management Inventory Data | Large-scale forest resource maps based on national forest inventory (NFI)
data and airborne laser scanning may facilitate synergies between NFIs and
forest management inventories (FMIs). A comparison of models used in such a
NFI-based map and a FMI indicate that NFI-based maps can directly be used in
FMIs to estimate timber volume of mature spruce forests. Traditionally, FMIs
and NFIs have been separate activities. The increasing availability of detailed
NFI-based forest resource maps provides the possibility to eliminate or reduce
the need of field sample plot measurements in FMIs if their accuracy is
similar. We aim to 1) compare a timber volume model used in a NFI-based map and
models used in a FMI, and 2) evaluate utilizing additional local sample plots
in the model of the NFI-based map. Accuracies of timber volume estimates using
models from an existing NFI-based map and a FMI were compared at plot and stand
level. Estimates from the NFI-based map were similar to or more accurate than
the FMI. The addition of local plots to the modeling data did not clearly
improve the model of the NFI-based map.The comparison indicates that NFI-based
maps can directly be used in FMIs for timber volume estimation in mature spruce
stands, leading to potentially large cost savings.
| stat |
Application of Deep Learning-based Interpolation Methods to Nearshore
Bathymetry | Nearshore bathymetry, the topography of the ocean floor in coastal zones, is
vital for predicting the surf zone hydrodynamics and for route planning to
avoid subsurface features. Hence, it is increasingly important for a wide
variety of applications, including shipping operations, coastal management, and
risk assessment. However, direct high resolution surveys of nearshore
bathymetry are rarely performed due to budget constraints and logistical
restrictions. Another option when only sparse observations are available is to
use Gaussian Process regression (GPR), also called Kriging. But GPR has
difficulties recognizing patterns with sharp gradients, like those found around
sand bars and submerged objects, especially when observations are sparse. In
this work, we present several deep learning-based techniques to estimate
nearshore bathymetry with sparse, multi-scale measurements. We propose a Deep
Neural Network (DNN) to compute posterior estimates of the nearshore
bathymetry, as well as a conditional Generative Adversarial Network (cGAN) that
samples from the posterior distribution. We train our neural networks based on
synthetic data generated from nearshore surveys provided by the U.S.\ Army
Corps of Engineer Field Research Facility (FRF) in Duck, North Carolina. We
compare our methods with Kriging on real surveys as well as surveys with
artificially added sharp gradients. Results show that direct estimation by DNN
gives better predictions than Kriging in this application. We use bootstrapping
with DNN for uncertainty quantification. We also propose a method, named
DNN-Kriging, that combines deep learning with Kriging and shows further
improvement of the posterior estimates.
| stat |
Multi-scale graph principal component analysis for connectomics | In brain connectomics, the cortical surface is parcellated into different
regions of interest (ROIs) prior to statistical analysis. The brain connectome
for each individual can then be represented as a graph, with the nodes
corresponding to ROIs and edges to connections between ROIs. Such a graph can
be summarized as an adjacency matrix, with each cell containing the strength of
connection between a pair of ROIs. These matrices are symmetric with the
diagonal elements corresponding to self-connections typically excluded. A major
disadvantage of such representations of the connectome is their sensitivity to
the chosen ROIs, including critically the number of ROIs and hence the scale of
the graph. As the scale becomes finer and more ROIs are used, graphs become
increasingly sparse. Clearly, the results of downstream statistical analyses
can be highly dependent on the chosen parcellation. To solve this problem, we
propose a multi-scale graph factorization, which links together scale-specific
factorizations through a common set of individual-specific scores. These scores
summarize an individual's brain structure combining information across
measurement scales. We obtain a simple and efficient algorithm for
implementation, and illustrate substantial advantages over single scale
approaches in simulations and analyses of the Human Connectome Project dataset.
| stat |
Comparison of statistical post-processing methods for probabilistic NWP
forecasts of solar radiation | The increased usage of solar energy places additional importance on forecasts
of solar radiation. Solar panel power production is primarily driven by the
amount of solar radiation and it is therefore important to have accurate
forecasts of solar radiation. Accurate forecasts that also give information on
the forecast uncertainties can help users of solar energy to make better solar
radiation based decisions related to the stability of the electrical grid. To
achieve this, we apply statistical post-processing techniques that determine
relationships between observations of global radiation (made within the KNMI
network of automatic weather stations in the Netherlands) and forecasts of
various meteorological variables from the numerical weather prediction (NWP)
model HARMONIE-AROME (HA) and the atmospheric composition model CAMS. Those
relationships are used to produce probabilistic forecasts of global radiation.
We compare 7 different statistical post-processing methods, consisting of two
parametric and five non-parametric methods. We find that all methods are able
to generate probabilistic forecasts that improve the raw global radiation
forecast from HA according to the root mean squared error (on the median) and
the potential economic value. Additionally, we show how important the
predictors are in the different regression methods. We also compare the
regression methods using various probabilistic scoring metrics, namely the
continuous ranked probability skill score, the Brier skill score and
reliability diagrams. We find that quantile regression and generalized random
forests generally perform best. In (near) clear sky conditions the
non-parametric methods have more skill than the parametric ones.
| stat |
Horseshoe Regularization for Machine Learning in Complex and Deep Models | Since the advent of the horseshoe priors for regularization, global-local
shrinkage methods have proved to be a fertile ground for the development of
Bayesian methodology in machine learning, specifically for high-dimensional
regression and classification problems. They have achieved remarkable success
in computation, and enjoy strong theoretical support. Most of the existing
literature has focused on the linear Gaussian case; see Bhadra et al. (2019b)
for a systematic survey. The purpose of the current article is to demonstrate
that the horseshoe regularization is useful far more broadly, by reviewing both
methodological and computational developments in complex models that are more
relevant to machine learning applications. Specifically, we focus on
methodological challenges in horseshoe regularization in nonlinear and
non-Gaussian models; multivariate models; and deep neural networks. We also
outline the recent computational developments in horseshoe shrinkage for
complex models along with a list of available software implementations that
allows one to venture out beyond the comfort zone of the canonical linear
regression problems.
| stat |
A generalised OMP algorithm for feature selection with application to
gene expression data | Feature selection for predictive analytics is the problem of identifying a
minimal-size subset of features that is maximally predictive of an outcome of
interest. To apply to molecular data, feature selection algorithms need to be
scalable to tens of thousands of available features. In this paper, we propose
gOMP, a highly-scalable generalisation of the Orthogonal Matching Pursuit
feature selection algorithm to several directions: (a) different types of
outcomes, such as continuous, binary, nominal, and time-to-event, (b) different
types of predictive models (e.g., linear least squares, logistic regression),
(c) different types of predictive features (continuous, categorical), and (d)
different, statistical-based stopping criteria. We compare the proposed
algorithm against LASSO, a prototypical, widely used algorithm for
high-dimensional data. On dozens of simulated datasets, as well as, real gene
expression datasets, gOMP is on par, or outperforms LASSO for case-control
binary classification, quantified outcomes (regression), and (censored)
survival times (time-to-event) analysis. gOMP has also several theoretical
advantages that are discussed. While gOMP is based on quite simple and basic
statistical ideas, easy to implement and to generalize, we also show in an
extensive evaluation that it is also quite effective in bioinformatics analysis
settings.
| stat |
Predicting the outputs of finite networks trained with noisy gradients | A recent line of works studied wide deep neural networks (DNNs) by
approximating them as Gaussian Processes (GPs). A DNN trained with gradient
flow was shown to map to a GP governed by the Neural Tangent Kernel (NTK),
whereas earlier works showed that a DNN with an i.i.d. prior over its weights
maps to the so-called Neural Network Gaussian Process (NNGP). Here we consider
a DNN training protocol, involving noise, weight decay and finite width, whose
outcome corresponds to a certain non-Gaussian stochastic process. An analytical
framework is then introduced to analyze this non-Gaussian process, whose
deviation from a GP is controlled by the finite width. Our contribution is
three-fold: (i) In the infinite width limit, we establish a correspondence
between DNNs trained with noisy gradients and the NNGP, not the NTK. (ii) We
provide a general analytical form for the finite width correction (FWC) for
DNNs with arbitrary activation functions and depth and use it to predict the
outputs of empirical finite networks with high accuracy. Analyzing the FWC
behavior as a function of $n$, the training set size, we find that it is
negligible for both the very small $n$ regime, and, surprisingly, for the large
$n$ regime (where the GP error scales as $O(1/n)$). (iii) We flesh-out
algebraically how these FWCs can improve the performance of finite
convolutional neural networks (CNNs) relative to their GP counterparts on image
classification tasks.
| stat |
Ongoing Vaccine and Monoclonal Antibody HIV Prevention Efficacy Trials
and Considerations for Sequel Efficacy Trial Designs | Four randomized placebo-controlled efficacy trials of a candidate vaccine or
passively infused monoclonal antibody for prevention of HIV-1 infection are
underway (HVTN 702 in South African men and women; HVTN 705 in sub-Saharan
African women; HVTN 703/HPTN 081 in sub-Saharan African women; HVTN 704/HPTN
085 in U.S., Peruvian, Brazilian, and Swiss men or transgender persons who have
sex with men). Several challenges are posed to the optimal design of the sequel
efficacy trials, including: (1) how to account for the evolving mosaic of
effective prevention interventions that may be part of the trial design or
standard of prevention; (2) how to define viable and optimal sequel trial
designs depending on the primary efficacy results and secondary 'correlates of
protection' results of each of the ongoing trials; and (3) how to define the
primary objective of sequel efficacy trials if HIV-1 incidence is expected to
be very low in all study arms such that a standard trial design has a steep
opportunity cost. After summarizing the ongoing trials, I discuss statistical
science considerations for sequel efficacy trial designs, both generally and
specifically to each trial listed above. One conclusion is that the results of
'correlates of protection' analyses, which ascertain how different host
immunological markers and HIV-1 viral features impact HIV-1 risk and prevention
efficacy, have an important influence on sequel trial design. This influence is
especially relevant for the monoclonal antibody trials because of the focused
pre-trial hypothesis that potency and coverage of serum neutralization
constitutes a surrogate endpoint for HIV-1 infection... (see manuscript for the
full abstract)
| stat |
Bayesian generalized linear model for over and under dispersed counts | Bayesian models that can handle both over and under dispersed counts are rare
in the literature, perhaps because full probability distributions for dispersed
counts are rather difficult to construct. This note takes a first look at
Bayesian Conway-Maxwell-Poisson generalized linear models that can handle both
over and under dispersion yet retain the parsimony and interpretability of
classical count regression models. The focus is on providing an explicit
demonstration of Bayesian regression inferences for dispersed counts via a
Metropolis-Hastings algorithm. We illustrate the approach on two data analysis
examples and demonstrate some favourable frequentist properties via a
simulation study.
| stat |
A Parsimonious Tour of Bayesian Model Uncertainty | Modern statistical software and machine learning libraries are enabling
semi-automated statistical inference. Within this context, it appears easier
and easier to try and fit many models to the data at hand, reversing thereby
the Fisherian way of conducting science by collecting data after the scientific
hypothesis (and hence the model) has been determined. The renewed goal of the
statistician becomes to help the practitioner choose within such large and
heterogeneous families of models, a task known as model selection. The Bayesian
paradigm offers a systematized way of assessing this problem. This approach,
launched by Harold Jeffreys in his 1935 book Theory of Probability, has
witnessed a remarkable evolution in the last decades, that has brought about
several new theoretical and methodological advances. Some of these recent
developments are the focus of this survey, which tries to present a unifying
perspective on work carried out by different communities. In particular, we
focus on non-asymptotic out-of-sample performance of Bayesian model selection
and averaging techniques, and draw connections with penalized maximum
likelihood. We also describe recent extensions to wider classes of
probabilistic frameworks including high-dimensional, unidentifiable, or
likelihood-free models.
| stat |
Kaplan-Meier based tests for exponentiality in the presence of censoring | In this paper we test the composite hypothesis that lifetimes follow an
exponential distribution based on observed randomly right censored data.
Testing this hypothesis is complicated by the presence of this censoring, due
to the fact that not all lifetimes are observed. To account for this
complication, we propose modifications to tests based on the empirical
characteristic function and Laplace transform. In the full sample case these
empirical functions can be expressed as integrals with respect to the empirical
distribution function of the lifetimes. We propose replacing this estimate of
the distribution function by the Kaplan-Meier estimate. The resulting test
statistics can be expressed in easily calculable forms in terms of summations
of functionals of the observed data. Additionally, a general framework for
goodness-of-fit testing, in the presence of random right censoring, is
outlined. A Monte Carlo study is performed, the results of which indicate that
the newly modified tests generally outperform the existing tests. A practical
application, concerning initial remission times of leukemia patients, is
discussed along with some concluding remarks and avenues for future research.
| stat |
An Alternative Data-Driven Prediction Approach Based on Real Option
Theories | This paper presents a new prediction model for time series data by
integrating a time-varying Geometric Brownian Motion model with a pricing
mechanism used in financial engineering. Typical time series models such as
Auto-Regressive Integrated Moving Average assumes a linear correlation
structure in time series data. When a stochastic process is highly volatile,
such an assumption can be easily violated, leading to inaccurate predictions.
We develop a new prediction model that can flexibly characterize a time-varying
volatile process without assuming linearity. We formulate the prediction
problem as an optimization problem with unequal overestimation and
underestimation costs. Based on real option theories developed in finance, we
solve the optimization problem and obtain a predicted value, which can minimize
the expected prediction cost. We evaluate the proposed approach using multiple
datasets obtained from real-life applications including manufacturing, finance,
and environment. The numerical results demonstrate that the proposed model
shows competitive prediction capability, compared with alternative approaches.
| stat |
Lower-bounded proper losses for weakly supervised classification | This paper discusses the problem of weakly supervised learning of
classification, in which instances are given weak labels that are produced by
some label-corruption process. The goal is to derive conditions under which
loss functions for weak-label learning are proper and lower-bounded -- two
essential requirements for the losses used in class-probability estimation. To
this end, we derive a representation theorem for proper losses in supervised
learning, which dualizes the Savage representation. We use this theorem to
characterize proper weak-label losses and find a condition for them to be
lower-bounded. Based on these theoretical findings, we derive a novel
regularization scheme called generalized logit squeezing, which makes any
proper weak-label loss bounded from below, without losing properness.
Furthermore, we experimentally demonstrate the effectiveness of our proposed
approach, as compared to improper or unbounded losses. Those results highlight
the importance of properness and lower-boundedness. The code is publicly
available at https://github.com/yoshum/lower-bounded-proper-losses.
| stat |
Non-Reversible Parallel Tempering: a Scalable Highly Parallel MCMC
Scheme | Parallel tempering (PT) methods are a popular class of Markov chain Monte
Carlo schemes used to sample complex high-dimensional probability
distributions. They rely on a collection of $N$ interacting auxiliary chains
targeting tempered versions of the target distribution to improve the
exploration of the state-space. We provide here a new perspective on these
highly parallel algorithms and their tuning by identifying and formalizing a
sharp divide in the behaviour and performance of reversible versus
non-reversible PT schemes. We show theoretically and empirically that a class
of non-reversible PT methods dominates its reversible counterparts and identify
distinct scaling limits for the non-reversible and reversible schemes, the
former being a piecewise-deterministic Markov process and the latter a
diffusion. These results are exploited to identify the optimal annealing
schedule for non-reversible PT and to develop an iterative scheme approximating
this schedule. We provide a wide range of numerical examples supporting our
theoretical and methodological contributions. The proposed methodology is
applicable to sample from a distribution $\pi$ with a density $L$ with respect
to a reference distribution $\pi_0$ and compute the normalizing constant. A
typical use case is when $\pi_0$ is a prior distribution, $L$ a likelihood
function and $\pi$ the corresponding posterior.
| stat |
Pseudo-Marginal Hamiltonian Monte Carlo | Bayesian inference in the presence of an intractable likelihood function is
computationally challenging. When following a Markov chain Monte Carlo (MCMC)
approach to approximate the posterior distribution in this context, one
typically either uses MCMC schemes which target the joint posterior of the
parameters and some auxiliary latent variables, or pseudo-marginal
Metropolis--Hastings (MH) schemes. The latter mimic a MH algorithm targeting
the marginal posterior of the parameters by approximating unbiasedly the
intractable likelihood. However, in scenarios where the parameters and
auxiliary variables are strongly correlated under the posterior and/or this
posterior is multimodal, Gibbs sampling or Hamiltonian Monte Carlo (HMC) will
perform poorly and the pseudo-marginal MH algorithm, as any other MH scheme,
will be inefficient for high dimensional parameters. We propose here an
original MCMC algorithm, termed pseudo-marginal HMC, which combines the
advantages of both HMC and pseudo-marginal schemes. Specifically, the
pseudo-marginal HMC method is controlled by a precision parameter N,
controlling the approximation of the likelihood and, for any N, it samples the
marginal posterior of the parameters. Additionally, as N tends to infinity, its
sample trajectories and acceptance probability converge to those of an ideal,
but intractable, HMC algorithm which would have access to the marginal
posterior of parameters and its gradient. We demonstrate through experiments
that pseudo-marginal HMC can outperform significantly both standard HMC and
pseudo-marginal MH schemes.
| stat |
Estimating the Optimal Linear Combination of Biomarkers using
Spherically Constrained Optimization | In the context of a binary classification problem, the optimal linear
combination of continuous predictors can be estimated by maximizing an
empirical estimate of the area under the receiver operating characteristic
(ROC) curve (AUC). For multi-category responses, the optimal predictor
combination can similarly be obtained by maximization of the empirical
hypervolume under the manifold (HUM). This problem is particularly relevant to
medical research, where it may be of interest to diagnose a disease with
various subtypes or predict a multi-category outcome. Since the empirical HUM
is discontinuous, non-differentiable, and possibly multi-modal, solving this
maximization problem requires a global optimization technique. Estimation of
the optimal coefficient vector using existing global optimization techniques is
computationally expensive, becoming prohibitive as the number of predictors and
the number of outcome categories increases. We propose an efficient
derivative-free black-box optimization technique based on pattern search to
solve this problem. Through extensive simulation studies, we demonstrate that
the proposed method achieves better performance compared to existing methods
including the step-down algorithm. Finally, we illustrate the proposed method
to predict swallowing difficulty after radiation therapy for oropharyngeal
cancer based on radiation dose to various structures in the head and neck.
| stat |
Predictive Inference Based on Markov Chain Monte Carlo Output | In Bayesian inference, predictive distributions are typically in the form of
samples generated via Markov chain Monte Carlo (MCMC) or related algorithms. In
this paper, we conduct a systematic analysis of how to make and evaluate
probabilistic forecasts from such simulation output. Based on proper scoring
rules, we develop a notion of consistency that allows to assess the adequacy of
methods for estimating the stationary distribution underlying the simulation
output. We then provide asymptotic results that account for the salient
features of Bayesian posterior simulators, and derive conditions under which
choices from the literature satisfy our notion of consistency. Importantly,
these conditions depend on the scoring rule being used, such that the choices
of approximation method and scoring rule are intertwined. While the logarithmic
rule requires fairly stringent conditions, the continuous ranked probability
score (CRPS) yields consistent approximations under minimal assumptions. These
results are illustrated in a simulation study and an economic data example.
Overall, mixture-of-parameters approximations which exploit the parametric
structure of Bayesian models perform particularly well. Under the CRPS, the
empirical distribution function is a simple and appealing alternative option.
| stat |
Double-Robust Estimation in Difference-in-Differences with an
Application to Traffic Safety Evaluation | Difference-in-differences (DID) is a widely used approach for drawing causal
inference from observational panel data. Two common estimation strategies for
DID are outcome regression and propensity score weighting. In this paper,
motivated by a real application in traffic safety research, we propose a new
double-robust DID estimator that hybridizes regression and propensity score
weighting. We particularly focus on the case of discrete outcomes. We show that
the proposed double-robust estimator possesses the desirable large-sample
robustness property. We conduct a simulation study to examine its finite-sample
performance and compare with alternative methods. Our empirical results from a
Pennsylvania Department of Transportation data suggest that rumble strips are
marginally effective in reducing vehicle crashes.
| stat |
A reproducing kernel Hilbert space framework for functional data
classification | We encounter a bottleneck when we try to borrow the strength of classical
classifiers to classify functional data. The major issue is that functional
data are intrinsically infinite dimensional, thus classical classifiers cannot
be applied directly or have poor performance due to the curse of
dimensionality. To address this concern, we propose to project functional data
onto one specific direction, and then a distance-weighted discrimination DWD
classifier is built upon the projection score. The projection direction is
identified through minimizing an empirical risk function that contains the
particular loss function in a DWD classifier, over a reproducing kernel Hilbert
space. Hence our proposed classifier can avoid overfitting and enjoy appealing
properties of DWD classifiers. This framework is further extended to
accommodate functional data classification problems where scalar covariates are
involved. In contrast to previous work, we establish a non-asymptotic
estimation error bound on the relative misclassification rate. In finite sample
case, we demonstrate that the proposed classifiers compare favorably with some
commonly used functional classifiers in terms of prediction accuracy through
simulation studies and a real-world application.
| stat |
Splitting strategies for post-selection inference | We consider the problem of providing valid inference for a selected parameter
in a sparse regression setting. It is well known that classical regression
tools can be unreliable in this context due to the bias generated in the
selection step. Many approaches have been proposed in recent years to ensure
inferential validity. Here, we consider a simple alternative to data splitting
based on randomising the response vector, which allows for higher selection and
inferential power than the former and is applicable with an arbitrary selection
rule. We provide a theoretical and empirical comparison of both methods and
extend the randomisation approach to non-normal settings. Our investigations
show that the gain in power can be substantial.
| stat |
High-resolution Spatio-temporal Model for County-level COVID-19 Activity
in the U.S | We present an interpretable high-resolution spatio-temporal model to estimate
COVID-19 deaths together with confirmed cases one-week ahead of the current
time, at the county-level and weekly aggregated, in the United States. A
notable feature of our spatio-temporal model is that it considers the (a)
temporal auto- and pairwise correlation of the two local time series (confirmed
cases and death of the COVID-19), (b) dynamics between locations (propagation
between counties), and (c) covariates such as local within-community mobility
and social demographic factors. The within-community mobility and demographic
factors, such as total population and the proportion of the elderly, are
included as important predictors since they are hypothesized to be important in
determining the dynamics of COVID-19. To reduce the model's
high-dimensionality, we impose sparsity structures as constraints and emphasize
the impact of the top ten metropolitan areas in the nation, which we refer (and
treat within our models) as hubs in spreading the disease. Our retrospective
out-of-sample county-level predictions were able to forecast the subsequently
observed COVID-19 activity accurately. The proposed multi-variate predictive
models were designed to be highly interpretable, with clear identification and
quantification of the most important factors that determine the dynamics of
COVID-19. Ongoing work involves incorporating more covariates, such as
education and income, to improve prediction accuracy and model
interpretability.
| stat |
Pixelate to communicate: visualising uncertainty in maps of disease risk
and other spatial continua | Maps have long been been used to visualise estimates of spatial variables, in
particular disease burden and risk. Predictions made using a geostatistical
model have uncertainty that typically varies spatially. However, this
uncertainty is difficult to map with the estimate itself and is often not
included as a result, thereby generating a potentially misleading sense of
certainty about disease burden or other important variables. To remedy this, we
propose simultaneously visualising predictions and their associated uncertainty
within a single map by varying pixel size. We illustrate our approach using
examples of malaria incidence, but the method could be applied to predictions
of any spatial continua with associated uncertainty.
| stat |
DeepHoops: Evaluating Micro-Actions in Basketball Using Deep Feature
Representations of Spatio-Temporal Data | How much is an on-ball screen worth? How much is a backdoor cut away from the
ball worth? Basketball is one of a number of sports which, within the past
decade, have seen an explosion in quantitative metrics and methods for
evaluating players and teams. However, it is still challenging to evaluate
individual off-ball events in terms of how they contribute to the success of a
possession. In this study, we develop an end-to-end deep learning architecture
DeepHoops to process a unique dataset composed of spatio-temporal tracking data
from NBA games in order to generate a running stream of predictions on the
expected points to be scored as a possession progresses. We frame the problem
as a multi-class sequence classification problem in which our model estimates
probabilities of terminal actions taken by players (e.g. take field goal,
turnover, foul etc.) at each moment of a possession based on a sequence of ball
and player court locations preceding the said moment. Each of these terminal
actions is associated with an expected point value, which is used to estimate
the expected points to be scored. One of the challenges associated with this
problem is the high imbalance in the action classes. To solve this problem, we
parameterize a downsampling scheme for the training phase. We demonstrate that
DeepHoops is well-calibrated, estimating accurately the probabilities of each
terminal action and we further showcase the model's capability to evaluate
individual actions (potentially off-ball) within a possession that are not
captured by boxscore statistics.
| stat |
Parsimony in Model Selection: Tools for Assessing Fit Propensity | Theories can be represented as statistical models for empirical testing.
There is a vast literature on model selection and multimodel inference that
focuses on how to assess which statistical model, and therefore which theory,
best fits the available data. For example, given some data, one can compare
models on various information criterion or other fit statistics. However, what
these indices fail to capture is the full range of counterfactuals. That is,
some models may fit the given data better not because they represent a more
correct theory, but simply because these models have more fit propensity - a
tendency to fit a wider range of data, even nonsensical data, better. Current
approaches fall short in considering the principle of parsimony (Occam's
Razor), often equating it with the number of model parameters. Here we offer a
toolkit for researchers to better study and understand parsimony through the
fit propensity of Structural Equation Models. We provide an R package
(ockhamSEM) built on the popular lavaan package. To illustrate the importance
of evaluating fit propensity, we use ockhamSEM to investigate the factor
structure of the Rosenberg Self-Esteem Scale.
| stat |
A Two-Stage Variable Selection Approach for Correlated High Dimensional
Predictors | When fitting statistical models, some predictors are often found to be
correlated with each other, and functioning together. Many group variable
selection methods are developed to select the groups of predictors that are
closely related to the continuous or categorical response. These existing
methods usually assume the group structures are well known. For example,
variables with similar practical meaning, or dummy variables created by
categorical data. However, in practice, it is impractical to know the exact
group structure, especially when the variable dimensional is large. As a
result, the group variable selection results may be selected. To solve the
challenge, we propose a two-stage approach that combines a variable clustering
stage and a group variable stage for the group variable selection problem. The
variable clustering stage uses information from the data to find a group
structure, which improves the performance of the existing group variable
selection methods. For ultrahigh dimensional data, where the predictors are
much larger than observations, we incorporated a variable screening method in
the first stage and shows the advantages of such an approach. In this article,
we compared and discussed the performance of four existing group variable
selection methods under different simulation models, with and without the
variable clustering stage. The two-stage method shows a better performance, in
terms of the prediction accuracy, as well as in the accuracy to select active
predictors. An athlete's data is also used to show the advantages of the
proposed method.
| stat |
An Empirical Study of Invariant Risk Minimization | Invariant risk minimization (IRM) (Arjovsky et al., 2019) is a recently
proposed framework designed for learning predictors that are invariant to
spurious correlations across different training environments. Yet, despite its
theoretical justifications, IRM has not been extensively tested across various
settings. In an attempt to gain a better understanding of the framework, we
empirically investigate several research questions using IRMv1, which is the
first practical algorithm proposed to approximately solve IRM. By extending the
ColoredMNIST experiment in different ways, we find that IRMv1 (i) performs
better as the spurious correlation varies more widely between training
environments, (ii) learns an approximately invariant predictor when the
underlying relationship is approximately invariant, and (iii) can be extended
to an analogous setting for text classification.
| stat |
Designing efficient randomized trials: power and sample size calculation
when using semiparametric efficient estimators | Trials enroll a large number of subjects in order to attain power, making
them expensive and time-consuming. Sample size calculations are often performed
with the assumption of an unadjusted analysis, even if the trial analysis plan
specifies a more efficient estimator (e.g. ANCOVA). This leads to conservative
estimates of required sample sizes and an opportunity for savings. Here we show
that a relatively simple formula can be used to estimate the power of any
two-arm, single-timepoint trial analyzed with a semiparametric efficient
estimator, regardless of the domain of the outcome or kind of treatment effect
(e.g. odds ratio, mean difference). Since an efficient estimator attains the
minimum possible asymptotic variance, this allows for the design of trials that
are as small as possible while still attaining design power and control of type
I error. The required sample size calculation is parsimonious and requires the
analyst to provide only a small number of population parameters. We verify in
simulation that the large-sample properties of trials designed this way attain
their nominal values. Lastly, we demonstrate how to use this formula in the
"design" (and subsequent reanalysis) of a real clinical trial and show that
fewer subjects are required to attain the same design power when a
semiparametric efficient estimator is accounted for at the design stage.
| stat |
Learning excursion sets of vector-valued Gaussian random fields for
autonomous ocean sampling | Improving and optimizing oceanographic sampling is a crucial task for marine
science and maritime resource management. Faced with limited resources in
understanding processes in the water-column, the combination of statistics and
autonomous systems provide new opportunities for experimental design. In this
work we develop efficient spatial sampling methods for characterizing regions
defined by simultaneous exceedances above prescribed thresholds of several
responses, with an application focus on mapping coastal ocean phenomena based
on temperature and salinity measurements. Specifically, we define a design
criterion based on uncertainty in the excursions of vector-valued Gaussian
random fields, and derive tractable expressions for the expected integrated
Bernoulli variance reduction in such a framework. We demonstrate how this
criterion can be used to prioritize sampling efforts at locations that are
ambiguous, making exploration more effective. We use simulations to study and
compare properties of the considered approaches, followed by results from field
deployments with an autonomous underwater vehicle as part of a study mapping
the boundary of a river plume. The results demonstrate the potential of
combining statistical methods and robotic platforms to effectively inform and
execute data-driven environmental sampling.
| stat |
Efficient implementation of median bias reduction with applications to
general regression models | In numerous regular statistical models, median bias reduction (Kenne Pagui et
al., 2017) has proven to be a noteworthy improvement over maximum likelihood,
alternative to mean bias reduction. The estimator is obtained as solution to a
modified score equation ensuring smaller asymptotic median bias than the
maximum likelihood estimator. This paper provides a simplified algebraic form
of the adjustment term for general regular models. With the new formula, the
estimation procedure benefits from a considerable computational gain by
avoiding multiple summations and thus allows an efficient implementation. More
importantly, the new formulation allows to highlight how the median bias
reduction adjustment can be obtained by adding an extra term to the mean bias
reduction adjustment. Illustrations are provided through new applications of
median bias reduction to two regression models not belonging to the generalized
linear models class, extended beta regression and beta-binomial regression.
Mean bias reduction is also provided here for the latter model. Simulation
studies show remarkable componentwise median centering of the median bias
reduced estimator, while variability and coverage of related confidence
intervals are comparable with those of mean bias reduction. Moreover, empirical
results for the beta-binomial model show that the method is successful in
solving maximum likelihood boundary estimate problem.
| stat |
A Tropical Approach to Neural Networks with Piecewise Linear Activations | We present a new, unifying approach following some recent developments on the
complexity of neural networks with piecewise linear activations. We treat
neural network layers with piecewise linear activations as tropical
polynomials, which generalize polynomials in the so-called $(\max, +)$ or
tropical algebra, with possibly real-valued exponents. Motivated by the
discussion in (arXiv:1402.1869), this approach enables us to refine their upper
bounds on linear regions of layers with ReLU or leaky ReLU activations to
$\min\left\{ 2^m, \sum_{j=0}^n \binom{m}{j} \right\}$, where $n, m$ are the
number of inputs and outputs, respectively. Additionally, we recover their
upper bounds on maxout layers. Our work follows a novel path, exclusively under
the lens of tropical geometry, which is independent of the improvements
reported in (arXiv:1611.01491, arXiv:1711.02114). Finally, we present a
geometric approach for effective counting of linear regions using random
sampling in order to avoid the computational overhead of exact counting
approaches
| stat |
Grouped Heterogeneous Mixture Modeling for Clustered Data | Clustered data is ubiquitous in a variety of scientific fields. In this
paper, we propose a flexible and interpretable modeling approach, called
grouped heterogenous mixture modeling, for clustered data, which models
cluster-wise conditional distributions by mixtures of latent conditional
distributions common to all the clusters. In the model, we assume that clusters
are divided into a finite number of groups and mixing proportions are the same
within the same group. We provide a simple generalized EM algorithm for
computing the maximum likelihood estimator, and an information criterion to
select the numbers of groups and latent distributions. We also propose
structured grouping strategies by introducing penalties on grouping parameters
in the likelihood function. Under the settings where both the number of
clusters and cluster sizes tend to infinity, we present asymptotic properties
of the maximum likelihood estimator and the information criterion. We
demonstrate the proposed method through simulation studies and an application
to crime risk modeling in Tokyo.
| stat |
Bayesian geoacoustic inversion using mixture density network | Bayesian geoacoustic inversion problems are conventionally solved by Markov
chain Monte Carlo methods or its variants, which are computationally expensive.
This paper extends the classic Bayesian geoacoustic inversion framework by
deriving important geoacoustic statistics of Bayesian geoacoustic inversion
from the multidimensional posterior probability density (PPD) using the mixture
density network (MDN) theory. These statistics make it convenient to train the
network directly on the whole parameter space and get the multidimensional PPD
of model parameters. The present approach provides a much more efficient way to
solve geoacoustic inversion problems in Bayesian inference framework. The
network is trained on a simulated dataset of surface-wave dispersion curves
with shear-wave velocities as labels and tested on both synthetic and real data
cases. The results show that the network gives reliable predictions and has
good generalization performance on unseen data. Once trained, the network can
rapidly (within seconds) give a fully probabilistic solution which is
comparable to Monte Carlo methods. It provides an promising approach for
real-time inversion.
| stat |
Optimal design of experiments to identify latent behavioral types | Bayesian optimal experiments that maximize the information gained from
collected data are critical to efficiently identify behavioral models. We
extend a seminal method for designing Bayesian optimal experiments by
introducing two computational improvements that make the procedure tractable:
(1) a search algorithm from artificial intelligence that efficiently explores
the space of possible design parameters, and (2) a sampling procedure which
evaluates each design parameter combination more efficiently. We apply our
procedure to a game of imperfect information to evaluate and quantify the
computational improvements. We then collect data across five different
experimental designs to compare the ability of the optimal experimental design
to discriminate among competing behavioral models against the experimental
designs chosen by a "wisdom of experts" prediction experiment. We find that
data from the experiment suggested by the optimal design approach requires
significantly less data to distinguish behavioral models (i.e., test
hypotheses) than data from the experiment suggested by experts. Substantively,
we find that reinforcement learning best explains human decision-making in the
imperfect information game and that behavior is not adequately described by the
Bayesian Nash equilibrium. Our procedure is general and computationally
efficient and can be applied to dynamically optimize online experiments.
| stat |
Adaptive Function-on-Scalar Regression with a Smoothing Elastic Net | This paper presents a new methodology, called AFSSEN, to simultaneously
select significant predictors and produce smooth estimates in a
high-dimensional function-on-scalar linear model with a sub-Gaussian errors.
Outcomes are assumed to lie in a general real separable Hilbert space, H, while
parameters lie in a subspace known as a Cameron Martin space, K, which are
closely related to Reproducing Kernel Hilbert Spaces, so that parameter
estimates inherit particular properties, such as smoothness or periodicity,
without enforcing such properties on the data. We propose a regularization
method in the style of an adaptive Elastic Net penalty that involves mixing two
types of functional norms, providing a fine tune control of both the smoothing
and variable selection in the estimated model. Asymptotic theory is provided in
the form of a functional oracle property, and the paper concludes with a
simulation study demonstrating the advantage of using AFSSEN over existing
methods in terms of prediction error and variable selection.
| stat |
Efficient generation of random derangements with the expected
distribution of cycle lengths | We show how to generate random derangements efficiently by two different
techniques: random restricted transpositions and sequential importance
sampling. The algorithm employing restricted transpositions can also be used to
generate random fixed-point-free involutions only, a.k.a. random perfect
matchings on the complete graph. Our data indicate that the algorithms generate
random samples with the expected distribution of cycle lengths, which we
derive, and for relatively small samples, which can actually be very large in
absolute numbers, we argue that they generate samples indistinguishable from
the uniform distribution. Both algorithms are simple to understand and
implement and possess a performance comparable to or better than those of
currently known methods. Simulations suggest that the mixing time of the
algorithm based on random restricted transpositions (in the total variance
distance with respect to the distribution of cycle lengths) is
$O(n^{a}\log{n}^{2})$ with $a \simeq \frac{1}{2}$ and $n$ the length of the
derangement. We prove that the sequential importance sampling algorithm
generates random derangements in $O(n)$ time with probability $O(1/n)$ of
failing.
| stat |
Gaussian mixture model decomposition of multivariate signals | We propose a greedy variational method for decomposing a non-negative
multivariate signal as a weighted sum of Gaussians, which, borrowing the
terminology from statistics, we refer to as a Gaussian mixture model. Notably,
our method has the following features: (1) It accepts multivariate signals,
i.e. sampled multivariate functions, histograms, time series, images, etc. as
input. (2) The method can handle general (i.e. ellipsoidal) Gaussians. (3) No
prior assumption on the number of mixture components is needed. To the best of
our knowledge, no previous method for Gaussian mixture model decomposition
simultaneously enjoys all these features. We also prove an upper bound, which
cannot be improved by a global constant, for the distance from any mode of a
Gaussian mixture model to the set of corresponding means. For mixtures of
spherical Gaussians with common variance $\sigma^2$, the bound takes the simple
form $\sqrt{n}\sigma$. We evaluate our method on one- and two-dimensional
signals. Finally, we discuss the relation between clustering and signal
decomposition, and compare our method to the baseline expectation maximization
algorithm.
| stat |
The Tajima heterochronous n-coalescent: inference from heterochronously
sampled molecular data | The observed sequence variation at a locus informs about the evolutionary
history of the sample and past population size dynamics. The standard Kingman
coalescent model on genealogies - timed trees that represent the ancestry of
the sample - is used in a generative model of molecular sequence variation to
infer evolutionary parameters. However, the state space of Kingman's
genealogies grows superexponentially with sample size n, making inference
computationally unfeasible already for small n. We introduce a new coalescent
model called Tajima heterochronous n-coalescent with a substantially smaller
cardinality of the genealogical space. This process allows to analyze samples
collected at different times, a situation that in applications is both met
(e.g. ancient DNA and RNA from rapidly evolving pathogens like viruses) and
statistically desirable (variance reduction and parameter identifiability). We
propose an algorithm to calculate the likelihood efficiently and present a
Bayesian nonparametric procedure to infer the population size trajectory. We
provide a new MCMC sampler to explore the space of Tajima's genealogies and
model parameters. We compare our procedure with state-of-the-art methodologies
in simulations and applications. We use our method to re-examine the scientific
question of how Beringian bison went extinct analyzing modern and ancient
molecular sequences of bison in North America, and to reconstruct population
size trajectory of SARS-CoV-2 from viral sequences collected in France and
Germany.
| stat |
A Bayesian spatio-temporal abundance model for surveillance of the
opioid epidemic | Opioid misuse is a national epidemic and a significant drug related threat to
the United States. While the scale of the problem is undeniable, estimates of
the local prevalence of opioid misuse are lacking, despite their importance to
policy-making and resource allocation. This is due, in part, to the challenge
of directly measuring opioid misuse at a local level. In this paper, we develop
a Bayesian hierarchical spatio-temporal abundance model that integrates
indirect county-level data on opioid overdose deaths and treatment admissions
with state-level survey estimates on prevalence of opioid misuse to estimate
the latent county-level prevalence and counts of people who misuse opioids. A
simulation study shows that our joint model accurately recovers the latent
counts and prevalence and thus overcomes known limitations with identifiability
in abundance models with non-replicated observations. We apply our model to
county-level surveillance data from the state of Ohio. Our proposed framework
can be applied to other applications of small area estimation for hard to reach
populations, which is a common occurrence with many health conditions such as
those related to illicit behaviors.
| stat |
An\'alisis estad\'istico ex post del conteo r\'apido institucional de la
elecci\'on de gobernador del Estado de M\'exico en 2017 | A statistical analysis of an electoral quick count based on the total count
of votes in the election of the State of Mexico's governor in 2017 is performed
in order to verify precision, confidence level of interval estimations,
possible bias and derived conclusions therein, with the main purpose of
checking compliance with the objectives of such statistical procedure.
-----
Se realiza un an\'alisis estad\'istico de las estimaciones del conteo
r\'apido institucional desde la perspectiva ideal de los resultados de los
c\'omputos distritales de la elecci\'on de gobernador del Estado de M\'exico
del a\~no 2017, particularmente aspectos como la precisi\'on de las
estimaciones, el nivel de confianza de los intervalos, el posible sesgo
respecto al c\'omputo distrital y las conclusiones que se derivaron y
reportaron, con el objetivo de determinar el grado de cumplimiento de los
objetivos de este ejercicio estad\'istico de car\'acter informativo.
| stat |
The MCC-F1 curve: a performance evaluation technique for binary
classification | Many fields use the ROC curve and the PR curve as standard evaluations of
binary classification methods. Analysis of ROC and PR, however, often gives
misleading and inflated performance evaluations, especially with an imbalanced
ground truth. Here, we demonstrate the problems with ROC and PR analysis
through simulations, and propose the MCC-F1 curve to address these drawbacks.
The MCC-F1 curve combines two informative single-threshold metrics, MCC and the
F1 score. The MCC-F1 curve more clearly differentiates good and bad
classifiers, even with imbalanced ground truths. We also introduce the MCC-F1
metric, which provides a single value that integrates many aspects of
classifier performance across the whole range of classification thresholds.
Finally, we provide an R package that plots MCC-F1 curves and calculates
related metrics.
| stat |
Causal Calculus in the Presence of Cycles, Latent Confounders and
Selection Bias | We prove the main rules of causal calculus (also called do-calculus) for i/o
structural causal models (ioSCMs), a generalization of a recently proposed
general class of non-/linear structural causal models that allow for cycles,
latent confounders and arbitrary probability distributions. We also generalize
adjustment criteria and formulas from the acyclic setting to the general one
(i.e. ioSCMs). Such criteria then allow to estimate (conditional) causal
effects from observational data that was (partially) gathered under selection
bias and cycles. This generalizes the backdoor criterion, the
selection-backdoor criterion and extensions of these to arbitrary ioSCMs.
Together, our results thus enable causal reasoning in the presence of cycles,
latent confounders and selection bias. Finally, we extend the ID algorithm for
the identification of causal effects to ioSCMs.
| stat |
Variational Laplace for Bayesian neural networks | We develop variational Laplace for Bayesian neural networks (BNNs) which
exploits a local approximation of the curvature of the likelihood to estimate
the ELBO without the need for stochastic sampling of the neural-network
weights. Variational Laplace performs better on image classification tasks than
MAP inference and far better than standard variational inference with
stochastic sampling despite using the same mean-field Gaussian approximate
posterior. The Variational Laplace objective is simple to evaluate, as it is
(in essence) the log-likelihood, plus weight-decay, plus a squared-gradient
regularizer. Finally, we emphasise care needed in benchmarking standard VI as
there is a risk of stopping before the variance parameters have converged. We
show that early-stopping can be avoided by increasing the learning rate for the
variance parameters.
| stat |
Power and sample size for cluster randomized and stepped wedge trials:
Comparing estimates obtained by applying design effects or by direct
estimation in GLMM | When observations are independent, formulae and software are readily
available to plan and design studies of appropriate size and power to detect
important associations. When observations are correlated or clustered, results
obtained from the standard software require adjustment. This tutorial compares
two approaches, using examples that illustrate various designs for both
independent and clustered data.
One approach obtains initial estimates using software that assume
independence among observations, then adjusts these estimates using a design
effect (DE), also called a variance inflation factor (VIF). A second approach
generates estimates using generalized linear mixed models (GLMM) that account
directly for patterns of clustering and correlation.
The two approaches generally produce similar estimates and so validate one
another. For certain clustered designs, small differences in power estimates
emphasize the importance of specifying an alternative hypothesis in terms of
means but also in terms of expected variances and covariances. Both approaches
to power estimation are sensitive to assumptions concerning the structure or
pattern of independence or correlation among clustered outcomes.
| stat |
Evaluating A Key Instrumental Variable Assumption Using Randomization
Tests | Instrumental variable (IV) analyses are becoming common in health services
research and epidemiology. Most IV analyses use naturally occurring
instruments, such as distance to a hospital. In these analyses, investigators
must assume the instrument is as-if randomly assigned. This assumption cannot
be tested directly, but it can be falsified. Most falsification tests in the
literature compare relative prevalence or bias in observed covariates between
the instrument and the exposure. These tests require investigators to make a
covariate-by-covariate judgment about the validity of the IV design. Often,
only some of the covariates are well-balanced, making it unclear if as-if
randomization can be assumed for the instrument across all covariates. We
propose an alternative falsification test that compares IV balance or bias to
the balance or bias that would have been produced under randomization. A key
advantage of our test is that it allows for global balance measures as well as
easily interpretable graphical comparisons. Furthermore, our test does not rely
on any parametric assumptions and can be used to validly assess if the
instrument is significantly closer to being as-if randomized than the exposure.
We demonstrate our approach on a recent IV application that uses bed
availability in the intensive care unit (ICU) as an instrument for admission to
the ICU.
| stat |
Neural Program Synthesis with a Differentiable Fixer | We present a new program synthesis approach that combines an encoder-decoder
based synthesis architecture with a differentiable program fixer. Our approach
is inspired from the fact that human developers seldom get their program
correct on the first attempt, and perform iterative testing-based program
fixing to get to the desired program functionality. Similarly, our approach
first learns a distribution over programs conditioned on an encoding of a set
of input-output examples, and then iteratively performs fix operations using
the differentiable fixer. The fixer takes as input the original examples and
the current program's outputs on example inputs, and generates a new
distribution over the programs with the goal of reducing the discrepancies
between the current program outputs and the desired example outputs. We train
our architecture end-to-end on the RobustFill domain, and show that the
addition of the fixer module leads to a significant improvement on synthesis
accuracy compared to using beam search.
| stat |
Effective Data Augmentation with Multi-Domain Learning GANs | For deep learning applications, the massive data development (e.g.,
collecting, labeling), which is an essential process in building practical
applications, still incurs seriously high costs. In this work, we propose an
effective data augmentation method based on generative adversarial networks
(GANs), called Domain Fusion. Our key idea is to import the knowledge contained
in an outer dataset to a target model by using a multi-domain learning GAN. The
multi-domain learning GAN simultaneously learns the outer and target dataset
and generates new samples for the target tasks. The simultaneous learning
process makes GANs generate the target samples with high fidelity and variety.
As a result, we can obtain accurate models for the target tasks by using these
generated samples even if we only have an extremely low volume target dataset.
We experimentally evaluate the advantages of Domain Fusion in image
classification tasks on 3 target datasets: CIFAR-100, FGVC-Aircraft, and Indoor
Scene Recognition. When trained on each target dataset reduced the samples to
5,000 images, Domain Fusion achieves better classification accuracy than the
data augmentation using fine-tuned GANs. Furthermore, we show that Domain
Fusion improves the quality of generated samples, and the improvements can
contribute to higher accuracy.
| stat |
Missing Not at Random in Matrix Completion: The Effectiveness of
Estimating Missingness Probabilities Under a Low Nuclear Norm Assumption | Matrix completion is often applied to data with entries missing not at random
(MNAR). For example, consider a recommendation system where users tend to only
reveal ratings for items they like. In this case, a matrix completion method
that relies on entries being revealed at uniformly sampled row and column
indices can yield overly optimistic predictions of unseen user ratings.
Recently, various papers have shown that we can reduce this bias in MNAR matrix
completion if we know the probabilities of different matrix entries being
missing. These probabilities are typically modeled using logistic regression or
naive Bayes, which make strong assumptions and lack guarantees on the accuracy
of the estimated probabilities. In this paper, we suggest a simple approach to
estimating these probabilities that avoids these shortcomings. Our approach
follows from the observation that missingness patterns in real data often
exhibit low nuclear norm structure. We can then estimate the missingness
probabilities by feeding the (always fully-observed) binary matrix specifying
which entries are revealed or missing to an existing nuclear-norm-constrained
matrix completion algorithm by Davenport et al. [2014]. Thus, we tackle MNAR
matrix completion by solving a different matrix completion problem first that
recovers missingness probabilities. We establish finite-sample error bounds for
how accurate these probability estimates are and how well these estimates
debias standard matrix completion losses for the original matrix to be
completed. Our experiments show that the proposed debiasing strategy can
improve a variety of existing matrix completion algorithms, and achieves
downstream matrix completion accuracy at least as good as logistic regression
and naive Bayes debiasing baselines that require additional auxiliary
information.
| stat |
Tensor Estimation with Nearly Linear Samples | There is a conjectured computational-statistical gap in terms of the number
of samples needed to perform tensor estimation. In particular, for a low rank
3-order tensor with $\Theta(n)$ parameters, Barak and Moitra conjectured that
$\Omega(n^{3/2})$ samples are needed for polynomial time computation based on a
reduction of a specific hard instance of a rank 1 tensor to the random 3-XOR
distinguishability problem. In this paper, we take a complementary perspective
and characterize a subclass of tensor instances that can be estimated with only
$O(n^{1+\kappa})$ observations for any arbitrarily small constant $\kappa > 0$,
nearly linear. If one considers the class of tensors with constant orthogonal
CP-rank, the "hardness" of the instance can be parameterized by the minimum
absolute value of the sum of latent factor vectors. If the sum of each latent
factor vector is bounded away from zero, we present an algorithm that can
perform tensor estimation with $O(n^{1+\kappa})$ samples for a $t$-order
tensor, significantly less than the previous achievable bound of $O(n^{t/2})$,
and close to the lower bound of $\Omega(n)$. This result suggests that amongst
constant orthogonal CP-rank tensors, the set of computationally hard instances
to estimate are in fact a small subset of all possible tensors.
| stat |
Generalizing trial findings using nested trial designs with sub-sampling
of non-randomized individuals | To generalize inferences from a randomized trial to the target population of
all trial-eligible individuals, investigators can use nested trial designs,
where the randomized individuals are nested within a cohort of trial-eligible
individuals, including those who are not offered or refuse randomization. In
these designs, data on baseline covariates are collected from the entire
cohort, and treatment and outcome data need only be collected from randomized
individuals. In this paper, we describe nested trial designs that improve
research economy by collecting additional baseline covariate data after
sub-sampling non-randomized individuals (i.e., a two-stage design), using
sampling probabilities that may depend on the initial set of baseline
covariates available from all individuals in the cohort. We propose an
estimator for the potential outcome mean in the target population of all
trial-eligible individuals and show that our estimator is doubly robust, in the
sense that it is consistent when either the model for the conditional outcome
mean among randomized individuals or the model for the probability of trial
participation is correctly specified. We assess the impact of sub-sampling on
the asymptotic variance of our estimator and examine the estimator's
finite-sample performance in a simulation study. We illustrate the methods
using data from the Coronary Artery Surgery Study (CASS).
| stat |
The Chi-Square Test of Distance Correlation | Distance correlation has gained much recent attention in the data science
community: the sample statistic is straightforward to compute and
asymptotically equals zero if and only if independence, making it an ideal
choice to test any type of dependency structure given sufficient sample size.
One major bottleneck is the testing process: because the null distribution of
distance correlation depends on the underlying random variables and metric
choice, it typically requires a permutation test to estimate the null and
compute the p-value, which is very costly for large amount of data. To overcome
the difficulty, we propose a centered chi-square distribution, demonstrate it
well-approximates the limiting null distribution of unbiased distance
correlation, and prove upper tail dominance and distribution bound. The
resulting distance correlation chi-square test is a nonparametric test for
independence, is valid and universally consistent using any strong negative
type metric or characteristic kernel, enjoys a similar finite-sample testing
power as the standard permutation test, is provably most powerful among all
valid tests of distance correlation using known distributions, and is also
applicable to K-sample and partial testing.
| stat |
Direct statistical inference for finite Markov jump processes via the
matrix exponential | Given noisy, partial observations of a time-homogeneous, finite-statespace
Markov chain, conceptually simple, direct statistical inference is available,
in theory, via its rate matrix, or infinitesimal generator, $\mathsf{Q}$, since
$\exp (\mathsf{Q}t)$ is the transition matrix over time $t$. However, perhaps
because of inadequate tools for matrix exponentiation in programming languages
commonly used amongst statisticians or a belief that the necessary calculations
are prohibitively expensive, statistical inference for continuous-time Markov
chains with a large but finite state space is typically conducted via particle
MCMC or other relatively complex inference schemes.
When, as in many applications $\mathsf{Q}$ arises from a reaction network, it
is usually sparse. We describe variations on known algorithms which allow fast,
robust and accurate evaluation of the product of a non-negative vector with the
exponential of a large, sparse rate matrix. Our implementation uses relatively
recently developed, efficient, linear algebra tools that take advantage of such
sparsity. We demonstrate the straightforward statistical application of the key
algorithm on a model for the mixing of two alleles in a population and on the
Susceptible-Infectious-Removed epidemic model.
| stat |
Measuring Within and Between Group Inequality in Early-Life Mortality
Over Time: A Bayesian Approach with Application to India | Most studies on inequality in infant and child mortality compare average
mortality rates between large groups of births, for example, comparing births
from different countries, income groups, ethnicities, or different times. These
studies do not measure within-group disparities. The few studies that have
measured within-group variability in infant and child mortality have used tools
from the income inequality literature, such as Gini indices. We show that the
latter are inappropriate for infant and child mortality. We develop novel tools
that are appropriate for analyzing infant and child mortality inequality,
including inequality measures, covariate adjustments, and ANOVA methods. We
illustrate how to handle uncertainty about complex inference targets, including
ensembles of probabilities and kernel density estimates. We illustrate our
methodology using a large data set from India, where we estimate infant and
child mortality risk for over 400,000 births using a Bayesian hierarchical
model. We show that most of the variance in mortality risk exists within groups
of births, not between them, and thus that within-group mortality needs to be
taken into account when assessing inequality in infant and child mortality. Our
approach has broad applicability to many health indicators.
| stat |
Modeling random and non-random decision uncertainty in ratings data: A
fuzzy beta model | Modeling human ratings data subject to raters' decision uncertainty is an
attractive problem in applied statistics. In view of the complex interplay
between emotion and decision making in rating processes, final raters' choices
seldom reflect the true underlying raters' responses. Rather, they are
imprecisely observed in the sense that they are subject to a non-random
component of uncertainty, namely the decision uncertainty. The purpose of this
article is to illustrate a statistical approach to analyse ratings data which
integrates both random and non-random components of the rating process. In
particular, beta fuzzy numbers are used to model raters' non-random decision
uncertainty and a variable dispersion beta linear model is instead adopted to
model the random counterpart of rating responses. The main idea is to quantify
characteristics of latent and non-fuzzy rating responses by means of random
observations subject to fuzziness. To do so, a fuzzy version of the
Expectation-Maximization algorithm is adopted to both estimate model's
parameters and compute their standard errors. Finally, the characteristics of
the proposed fuzzy beta model are investigated by means of a simulation study
as well as two case studies from behavioral and social contexts.
| stat |
A Bayesian Zero-Inflated Negative Binomial Regression Model for the
Integrative Analysis of Microbiome Data | Microbiome `omics approaches can reveal intriguing relationships between the
human microbiome and certain disease states. Along with the identification of
specific bacteria taxa associated with diseases, recent scientific advancements
provide mounting evidence that metabolism, genetics and environmental factors
can all modulate these microbial effects. However, the current methods for
integrating microbiome data and other covariates are severely lacking. Hence,
we present an integrative Bayesian zero-inflated negative binomial regression
model that can both distinguish differentially abundant taxa with distinct
phenotypes and quantify covariate-taxa effects. Our model demonstrates good
performance using simulated data. Furthermore, we successfully integrated
microbiome taxonomies and metabolomics in two real microbiome datasets to
provide biologically interpretable findings. In all, we proposed a novel
integrative Bayesian regression model that features bacterial differential
abundance analysis and microbiome-covariate effects quantifications, which
makes it suitable for general microbiome studies.
| stat |
Dynamics of stochastic gradient descent for two-layer neural networks in
the teacher-student setup | Deep neural networks achieve stellar generalisation even when they have
enough parameters to easily fit all their training data. We study this
phenomenon by analysing the dynamics and the performance of over-parameterised
two-layer neural networks in the teacher-student setup, where one network, the
student, is trained on data generated by another network, called the teacher.
We show how the dynamics of stochastic gradient descent (SGD) is captured by a
set of differential equations and prove that this description is asymptotically
exact in the limit of large inputs. Using this framework, we calculate the
final generalisation error of student networks that have more parameters than
their teachers. We find that the final generalisation error of the student
increases with network size when training only the first layer, but stays
constant or even decreases with size when training both layers. We show that
these different behaviours have their root in the different solutions SGD finds
for different activation functions. Our results indicate that achieving good
generalisation in neural networks goes beyond the properties of SGD alone and
depends on the interplay of at least the algorithm, the model architecture, and
the data set.
| stat |
Excess risk bounds in robust empirical risk minimization | This paper investigates robust versions of the general empirical risk
minimization algorithm, one of the core techniques underlying modern
statistical methods. Success of the empirical risk minimization is based on the
fact that for a "well-behaved" stochastic process $\left\{ f(X), \ f\in
\mathcal F\right\}$ indexed by a class of functions $f\in \mathcal F$, averages
$\frac{1}{N}\sum_{j=1}^N f(X_j)$ evaluated over a sample $X_1,\ldots,X_N$ of
i.i.d. copies of $X$ provide good approximation to the expectations $\mathbb E
f(X)$ uniformly over large classes $f\in \mathcal F$. However, this might no
longer be true if the marginal distributions of the process are heavy-tailed or
if the sample contains outliers. We propose a version of empirical risk
minimization based on the idea of replacing sample averages by robust proxies
of the expectation, and obtain high-confidence bounds for the excess risk of
resulting estimators. In particular, we show that the excess risk of robust
estimators can converge to $0$ at fast rates with respect to the sample size.
We discuss implications of the main results to the linear and logistic
regression problems, and evaluate the numerical performance of proposed methods
on simulated and real data.
| stat |
Application of the Cox Regression Model for Analysis of Railway Safety
Performance | The assessment of in-service safety performance is an important task, not
only in railways. For example it is important to identify deviations early, in
particular possible deterioration of safety performance, so that corrective
actions can be applied early. On the other hand the assessment should be fair
and objective and rely on sound and proven statistical methods. A popular means
for this task is trend analysis. This paper defines a model for trend analysis
and compares different approaches, e. g. classical and Bayes approaches, on
real data. The examples show that in particular for small sample sizes, e. g.
when railway operators shall be assessed, the Bayesian prior may influence the
results significantly.
| stat |
Privacy-preserving data sharing via probabilistic modelling | Differential privacy allows quantifying privacy loss resulting from accessing
sensitive personal data. Repeated accesses to underlying data incur increasing
loss. Releasing data as privacy-preserving synthetic data would avoid this
limitation, but would leave open the problem of designing what kind of
synthetic data. We propose formulating the problem of private data release
through probabilistic modelling. This approach transforms the problem of
designing the synthetic data into choosing a model for the data, allowing also
including prior knowledge, which improves the quality of the synthetic data. We
demonstrate empirically, in an epidemiological study, that statistical
discoveries can be reliably reproduced from the synthetic data. We expect the
method to have broad use in creating high-quality anonymized data twins of key
data sets for research.
| stat |
Technical efficiency and inefficiency: SFA misspecification | The effect of external factors $z$ on technical inefficiency ($TI$) in
stochastic frontier (SF) production models is often specified through the
variance of inefficiency term $u$. In this setup the signs of marginal effects
of $z$ on $TI$ and technical efficiency $TE$ identify how one should control
$z$ to increase $TI$ or decrease $TE$. We prove that these signs for $TI$ and
$TE$ are opposite for typical setups with normally distributed random error $v$
and exponentially or half-normally distributed $u$ for both conditional and
unconditional cases.
On the other hand, we give an example to show that signs of the marginal
effects of $z$ on $TI$ and $TE$ may coincide, at least for some ranges of $z$.
In our example, the distribution of $u$ is a mixture of two distributions, and
the proportion of the mixture is a function of $z$. Thus if the real data comes
from this mixture distribution, and we estimate model parameters with an
exponential or half-normal distribution for $u$, the estimated efficiency and
the marginal effect of $z$ on $TE$ would be wrong. Moreover, for a misspecified
model, the rank correlations between the true and the estimated values of TE
could be small and even negative for some subsamples of data. These results are
demonstrated by simulations.
| stat |
Subgroup-based Rank-1 Lattice Quasi-Monte Carlo | Quasi-Monte Carlo (QMC) is an essential tool for integral approximation,
Bayesian inference, and sampling for simulation in science, etc. In the QMC
area, the rank-1 lattice is important due to its simple operation, and nice
properties for point set construction. However, the construction of the
generating vector of the rank-1 lattice is usually time-consuming because of an
exhaustive computer search. To address this issue, we propose a simple
closed-form rank-1 lattice construction method based on group theory. Our
method reduces the number of distinct pairwise distance values to generate a
more regular lattice. We theoretically prove a lower and an upper bound of the
minimum pairwise distance of any non-degenerate rank-1 lattice. Empirically,
our methods can generate a near-optimal rank-1 lattice compared with the
Korobov exhaustive search regarding the $l_1$-norm and $l_2$-norm minimum
distance. Moreover, experimental results show that our method achieves superior
approximation performance on benchmark integration test problems and kernel
approximation problems.
| stat |
Variational Discriminant Analysis with Variable Selection | A fast Bayesian method that seamlessly fuses classification and hypothesis
testing via discriminant analysis is developed. Building upon the original
discriminant analysis classifier, modelling components are added to identify
discriminative variables. A combination of cake priors and a novel form of
variational Bayes we call reverse collapsed variational Bayes gives rise to
variable selection that can be directly posed as a multiple hypothesis testing
approach using likelihood ratio statistics. Some theoretical arguments are
presented showing that Chernoff-consistency (asymptotically zero type I and
type II error) is maintained across all hypotheses. We apply our method on some
publicly available genomics datasets and show that our method performs well in
practice for its computational cost. An R package VaDA has also been made
available on Github.
| stat |
Medium-Term Load Forecasting Using Support Vector Regression, Feature
Selection, and Symbiotic Organism Search Optimization | An accurate load forecasting has always been one of the main indispensable
parts in the operation and planning of power systems. Among different time
horizons of forecasting, while short-term load forecasting (STLF) and long-term
load forecasting (LTLF) have respectively got benefits of accurate predictors
and probabilistic forecasting, medium-term load forecasting (MTLF) demands more
attention due to its vital role in power system operation and planning such as
optimal scheduling of generation units, robust planning program for customer
service, and economic supply. In this study, a hybrid method, composed of
Support Vector Regression (SVR) and Symbiotic Organism Search Optimization
(SOSO) method, is proposed for MTLF. In the proposed forecasting model, SVR is
the main part of the forecasting algorithm while SOSO is embedded into it to
optimize the parameters of SVR. In addition, a minimum redundancy-maximum
relevance feature selection algorithm is used to in the preprocessing of input
data. The proposed method is tested on EUNITE competition dataset to
demonstrate its proper performance. Furthermore, it is compared with some
previous works to show eligibility of our method.
| stat |
A 1000-fold Acceleration of Hidden Markov Model Fitting using Graphical
Processing Units, with application to Nonvolcanic Tremor Classification | Hidden Markov models (HMMs) are general purpose models for time-series data
widely used across the sciences because of their flexibility and elegance.
However fitting HMMs can often be computationally demanding and time consuming,
particularly when the the number of hidden states is large or the Markov chain
itself is long. Here we introduce a new Graphical Processing Unit (GPU) based
algorithm designed to fit long chain HMMs, applying our approach to an HMM for
nonvolcanic tremor events developed by Wang et al.(2018). Even on a modest GPU,
our implementation resulted in a 1000-fold increase in speed over the standard
single processor algorithm, allowing a full Bayesian inference of uncertainty
related to model parameters. Similar improvements would be expected for HMM
models given large number of observations and moderate state spaces (<80 states
with current hardware). We discuss the model, general GPU architecture and
algorithms and report performance of the method on a tremor dataset from the
Shikoku region, Japan.
| stat |
BRIDGE: Byzantine-resilient Decentralized Gradient Descent | Decentralized optimization techniques are increasingly being used to learn
machine learning models from data distributed over multiple locations without
gathering the data at any one location. Unfortunately, methods that are
designed for faultless networks typically fail in the presence of node
failures. In particular, Byzantine failures---corresponding to the scenario in
which faulty/compromised nodes are allowed to arbitrarily deviate from an
agreed-upon protocol---are the hardest to safeguard against in decentralized
settings. This paper introduces a Byzantine-resilient decentralized gradient
descent (BRIDGE) method for decentralized learning that, when compared to
existing works, is more efficient and scalable in higher-dimensional settings
and that is deployable in networks having topologies that go beyond the star
topology. The main contributions of this work include theoretical analysis of
BRIDGE for strongly convex learning objectives and numerical experiments
demonstrating the efficacy of BRIDGE for both convex and nonconvex learning
tasks.
| stat |
The Age-Period-Cohort-Interaction Model for Describing and Investigating
Inter-Cohort Deviations and Intra-Cohort Life-Course Dynamics | Social scientists have frequently sought to understand the distinct effects
of age, period, and cohort, but disaggregation of the three dimensions is
difficult because cohort = period - age. We argue that this technical
difficulty reflects a disconnection between how cohort effect is conceptualized
and how it is modeled in the traditional age-period-cohort framework. We
propose a new method, called the age-period-cohort-interaction (APC-I) model,
that is qualitatively different from previous methods in that it represents
Ryder's (1965) theoretical account about the conditions under which cohort
differentiation may arise. This APC-I model does not require problematic
statistical assumptions and the interpretation is straightforward. It
quantifies inter-cohort deviations from the age and period main effects and
also permits hypothesis testing about intra-cohort life-course dynamics. We
demonstrate how this new model can be used to examine age, period, and cohort
patterns in women's labor force participation.
| stat |
What do AI algorithms actually learn? - On false structures in deep
learning | There are two big unsolved mathematical questions in artificial intelligence
(AI): (1) Why is deep learning so successful in classification problems and (2)
why are neural nets based on deep learning at the same time universally
unstable, where the instabilities make the networks vulnerable to adversarial
attacks. We present a solution to these questions that can be summed up in two
words; false structures. Indeed, deep learning does not learn the original
structures that humans use when recognising images (cats have whiskers, paws,
fur, pointy ears, etc), but rather different false structures that correlate
with the original structure and hence yield the success. However, the false
structure, unlike the original structure, is unstable. The false structure is
simpler than the original structure, hence easier to learn with less data and
the numerical algorithm used in the training will more easily converge to the
neural network that captures the false structure. We formally define the
concept of false structures and formulate the solution as a conjecture. Given
that trained neural networks always are computed with approximations, this
conjecture can only be established through a combination of theoretical and
computational results similar to how one establishes a postulate in theoretical
physics (e.g. the speed of light is constant). Establishing the conjecture
fully will require a vast research program characterising the false structures.
We provide the foundations for such a program establishing the existence of the
false structures in practice. Finally, we discuss the far reaching consequences
the existence of the false structures has on state-of-the-art AI and Smale's
18th problem.
| stat |
Estimation Methods for Cluster Randomized Trials with Noncompliance: A
Study of A Biometric Smartcard Payment System in India | Many policy evaluations occur in settings where treatment is randomized at
the cluster level, and there is treatment noncompliance within each cluster.
For example, villages might be assigned to treatment and control, but residents
in each village may choose to comply or not with their assigned treatment
status. When noncompliance is present, the instrumental variables framework can
be used to identify and estimate causal effects. While a large literature
exists on instrumental variables estimation methods, relatively little work has
been focused on settings with clustered treatments. Here, we review extant
methods for instrumental variable estimation in clustered designs and derive
both the finite and asymptotic properties of these estimators. We prove that
the properties of current estimators depend on unrealistic assumptions. We then
develop a new IV estimation method for cluster randomized trials and study its
formal properties. We prove that our IV estimator allows for possible treatment
effect heterogeneity that is correlated with cluster size and is robust to low
compliance rates within clusters. We evaluate these methods using simulations
and apply them to data from a randomized intervention in India.
| stat |
Bayesian Deconditional Kernel Mean Embeddings | Conditional kernel mean embeddings form an attractive nonparametric framework
for representing conditional means of functions, describing the observation
processes for many complex models. However, the recovery of the original
underlying function of interest whose conditional mean was observed is a
challenging inference task. We formalize deconditional kernel mean embeddings
as a solution to this inverse problem, and show that it can be naturally viewed
as a nonparametric Bayes' rule. Critically, we introduce the notion of task
transformed Gaussian processes and establish deconditional kernel means as
their posterior predictive mean. This connection provides Bayesian
interpretations and uncertainty estimates for deconditional kernel mean
embeddings, explains their regularization hyperparameters, and reveals a
marginal likelihood for kernel hyperparameter learning. These revelations
further enable practical applications such as likelihood-free inference and
learning sparse representations for big data.
| stat |
Estimation of a Low-rank Topic-Based Model for Information Cascades | We consider the problem of estimating the latent structure of a social
network based on the observed information diffusion events, or cascades, where
the observations for a given cascade consist of only the timestamps of
infection for infected nodes but not the source of the infection. Most of the
existing work on this problem has focused on estimating a diffusion matrix
without any structural assumptions on it. In this paper, we propose a novel
model based on the intuition that an information is more likely to propagate
among two nodes if they are interested in similar topics which are also
prominent in the information content. In particular, our model endows each node
with an influence vector (which measures how authoritative the node is on each
topic) and a receptivity vector (which measures how susceptible the node is for
each topic). We show how this node-topic structure can be estimated from the
observed cascades, and prove the consistency of the estimator. Experiments on
synthetic and real data demonstrate the improved performance and better
interpretability of our model compared to existing state-of-the-art methods.
| stat |
Optimal transport for vector Gaussian mixture models | Vector Gaussian mixture models form an important special subset of
vector-valued distributions. Any physical entity that can mutate or transit
among alternative manifestations distributed in a given space falls into this
category. A key example is color imagery. In this note, we vectorize the
Gaussian mixture model and study different optimal mass transport related
problems for such models. The benefits of using vector Gaussian mixture for
optimal mass transport include computational efficiency and the ability to
preserve structure.
| stat |
Comparison of plotting system outputs in beginner analysts | The R programming language is built on an ecosystem of packages, some that
allow analysts to accomplish the same tasks. For example, there are at least
two clear workflows for creating data visualizations in R: using the base
graphics package (referred to as "base R") and the ggplot2 add-on package based
on the grammar of graphics. Here we perform an empirical study of the quality
of scientific graphics produced by beginning R users. In our experiment,
learners taking a data science course on the Coursera platform were randomized
to complete identical plotting exercises in either the base R or the ggplot2
system. Learners were then asked to evaluate their peers in terms of visual
characteristics key to scientific cognition. We observed that graphics created
with the two systems rated similarly on many characteristics. However, ggplot2
graphics were generally judged to be more visually pleasing and, in the case of
faceted scientific plots, easier to understand. Our results suggest that while
both graphic systems are useful in the hands of beginning users, ggplot2's
natural faceting system may be easier to use by beginning users for displaying
more complex relationships.
| stat |
Making the most of imprecise measurements: Changing patterns of arsenic
concentrations in shallow wells of Bangladesh from laboratory and field data | Millions of people in Bangladesh drink well water contaminated with arsenic.
Despite the severity of this heath crisis, little is known about the extent to
which groundwater arsenic concentrations change over time: Are concentrations
generally rising, or is arsenic being flushed out of aquifers? Are spatially
patterns of high and low concentrations across wells homogenizing over time, or
are these spatial gradients becoming more pronounced? To address these
questions, we analyze a large set of arsenic concentrations that were sampled
within a 25 km$^2$ area of Bangladesh over time. We compare two blanket survey
collected in 2000/2001 and 2012/2013 from the same villages but relying on a
largely different set of wells. The early set consists of 4574 accurate
laboratory measurements, but the later set poses a challenge for analysis
because it is composed of 8229 less accurate categorical measurements conducted
in the field with a kit. We construct a Bayesian model that jointly calibrates
the measurement errors, applies spatial smoothing, and describes the
spatiotemporal dynamic with a diffusion-like process model. Our statistical
analysis reveals that arsenic concentrations change over time and that their
mean dropped from 110 to 96 $\mu$g/L over 12 years, although one quarter of
individual wells are inferred to see an increase. The largest decreases
occurred at the wells with locally high concentrations where the estimated
Laplacian indicated that the arsenic surface was strongly concave. However,
well with initially low concentrations were unlikely to be contaminated by
nearby high concentration wells over a decade. We validate the model using a
posterior predictive check on an external subset of laboratory measurements
from the same 271 wells in the same study area available for 2000, 2014, and
2015.
| stat |
A Continuous-Time Dynamic Choice Measurement Model for Problem-Solving
Process Data | Problem solving has been recognized as a central skill that today's students
need to thrive and shape their world. As a result, the measurement of
problem-solving competency has received much attention in education in recent
years. A popular tool for the measurement of problem solving is simulated
interactive tasks, which require students to uncover some of the information
needed to solve the problem through interactions with a computer-simulated
environment. A computer log file records a student's problem-solving process in
details, including his/her actions and the time stamps of these actions. It
thus provides rich information for the measurement of students' problem-solving
competency. On the other hand, extracting useful information from log files is
a challenging task, due to its complex data structure. In this paper, we show
how log file process data can be viewed as a marked point process, based on
which we propose a continuous-time dynamic choice model. The proposed model can
serve as a measurement model for scaling students along the latent traits of
problem-solving competency and action speed, based on data from one or multiple
tasks. A real data example is given based on data from Program for
International Student Assessment 2012.
| stat |
Whitening and Coloring batch transform for GANs | Batch Normalization (BN) is a common technique used to speed-up and stabilize
training. On the other hand, the learnable parameters of BN are commonly used
in conditional Generative Adversarial Networks (cGANs) for representing
class-specific information using conditional Batch Normalization (cBN). In this
paper we propose to generalize both BN and cBN using a Whitening and Coloring
based batch normalization. We show that our conditional Coloring can represent
categorical conditioning information which largely helps the cGAN qualitative
results. Moreover, we show that full-feature whitening is important in a
general GAN scenario in which the training process is known to be highly
unstable. We test our approach on different datasets and using different GAN
networks and training protocols, showing a consistent improvement in all the
tested frameworks. Our CIFAR-10 conditioned results are higher than all
previous works on this dataset.
| stat |
LinDA: Linear Models for Differential Abundance Analysis of Microbiome
Compositional Data | One fundamental statistical task in microbiome data analysis is differential
abundance analysis, which aims to identify microbial taxa whose abundance
covaries with a variable of interest. Although the main interest is on the
change in the absolute abundance, i.e., the number of microbial cells per unit
area/volume at the ecological site such as the human gut, the data from a
sequencing experiment reflects only the taxa relative abundances in a sample.
Thus, microbiome data are compositional in nature. Analysis of such
compositional data is challenging since the change in the absolute abundance of
one taxon will lead to changes in the relative abundances of other taxa, making
false positive control difficult. Here we present a simple, yet robust and
highly scalable approach to tackle the compositional effects in differential
abundance analysis. The method only requires the application of established
statistical tools. It fits linear regression models on the centered log-ratio
transformed data, identifies a bias term due to the transformation and
compositional effect, and corrects the bias using the mode of the regression
coefficients. Due to the algorithmic simplicity, our method is 100-1000 times
faster than the state-of-the-art method ANCOM-BC. Under mild assumptions, we
prove its asymptotic FDR control property, making it the first differential
abundance method that enjoys a theoretical FDR control guarantee. The proposed
method is very flexible and can be extended to mixed-effect models for the
analysis of correlated microbiome data. Using comprehensive simulations and
real data applications, we demonstrate that our method has overall the best
performance in terms of FDR control and power among the competitors. We
implemented the proposed method in the R package LinDA
(https://github.com/zhouhj1994/LinDA).
| stat |
Forecasting, Causality, and Impulse Response with Neural Vector
Autoregressions | Incorporating nonlinearity is paramount to predicting the future states of a
dynamical system, its response to shocks, and its underlying causal network.
However, most existing methods for causality detection and impulse response,
such as Vector Autoregression (VAR), assume linearity and are thus unable to
capture the complexity. Here, we introduce a vector autoencoder nonlinear
autoregression neural network (VANAR) capable of both automatic time series
feature extraction for its inputs and functional form estimation. We evaluate
VANAR in three ways: first in terms of pure forecast accuracy, second in terms
of detecting the correct causality between variables, and lastly in terms of
impulse response where we model trajectories given external shocks. These tests
were performed on a simulated nonlinear chaotic system and an empirical system
using Philippine macroeconomic data. Results show that VANAR significantly
outperforms VAR in the forecast and causality tests. VANAR has consistently
superior accuracy even over state of the art models such as SARIMA and TBATS.
For the impulse response test, both models fail to predict the shocked
trajectories of the nonlinear chaotic system. VANAR was robust in its ability
to model a wide variety of dynamics, from chaotic, high noise, and low data
environments to macroeconomic systems.
| stat |
Sparse Bayesian Causal Forests for Heterogeneous Treatment Effects
Estimation | This paper develops a sparsity-inducing version of Bayesian Causal Forests, a
recently proposed nonparametric causal regression model that employs Bayesian
Additive Regression Trees and is specifically designed to estimate
heterogeneous treatment effects using observational data. The sparsity-inducing
component we introduce is motivated by empirical studies where the number of
pre-treatment covariates available is non-negligible, leading to different
degrees of sparsity underlying the surfaces of interest in the estimation of
individual treatment effects. The extended version presented in this work,
which we name Sparse Bayesian Causal Forest, is equipped with an additional
pair of priors allowing the model to adjust the weight of each covariate
through the corresponding number of splits in the tree ensemble. These priors
improve the model's adaptability to sparse settings and allow to perform fully
Bayesian variable selection in a framework for treatment effects estimation,
and thus to uncover the moderating factors driving heterogeneity. In addition,
the method allows prior knowledge about the relevant confounding pre-treatment
covariates and the relative magnitude of their impact on the outcome to be
incorporated in the model. We illustrate the performance of our method in
simulated studies, in comparison to Bayesian Causal Forest and other
state-of-the-art models, to demonstrate how it scales up with an increasing
number of covariates and how it handles strongly confounded scenarios. Finally,
we also provide an example of application using real-world data.
| stat |
Post-processing numerical weather prediction ensembles for probabilistic
solar irradiance forecasting | In order to enable the transition towards renewable energy sources,
probabilistic energy forecasting is of critical importance for incorporating
volatile power sources such as solar energy into the electrical grid. Solar
energy forecasting methods often aim to provide probabilistic predictions of
solar irradiance. In particular, many hybrid approaches combine physical
information from numerical weather prediction models with statistical methods.
Even though the physical models can provide useful information at intra-day and
day-ahead forecast horizons, ensemble weather forecasts from multiple model
runs are often not calibrated and show systematic biases. We propose a
post-processing model for ensemble weather predictions of solar irradiance at
temporal resolutions between 30 minutes and 6 hours. The proposed models
provide probabilistic forecasts in the form of a censored logistic probability
distribution for lead times up to 5 days and are evaluated in two case studies
covering distinct physical models, geographical regions, temporal resolutions,
and types of solar irradiance. We find that post-processing consistently and
significantly improves the forecast performance of the ensemble predictions for
lead times up to at least 48 hours and is well able to correct the systematic
lack of calibration.
| stat |
CoolMomentum: A Method for Stochastic Optimization by Langevin Dynamics
with Simulated Annealing | Deep learning applications require optimization of nonconvex objective
functions. These functions have multiple local minima and their optimization is
a challenging problem. Simulated Annealing is a well-established method for
optimization of such functions, but its efficiency depends on the efficiency of
the adapted sampling methods. We explore relations between the Langevin
dynamics and stochastic optimization. By combining the Momentum optimizer with
Simulated Annealing, we propose CoolMomentum - a prospective stochastic
optimization method. Empirical results confirm the efficiency of the proposed
theoretical approach.
| stat |
Calibrated Top-1 Uncertainty estimates for classification by score based
models | While the accuracy of modern deep learning models has significantly improved
in recent years, the ability of these models to generate uncertainty estimates
has not progressed to the same degree. Uncertainty methods are designed to
provide an estimate of class probabilities when predicting class assignment.
While there are a number of proposed methods for estimating uncertainty, they
all suffer from a lack of calibration: predicted probabilities can be off from
empirical ones by a few percent or more. By restricting the scope of our
predictions to only the probability of Top-1 error, we can decrease the
calibration error of existing methods to less than one percent. As a result,
the scores of the methods also improve significantly over benchmarks.
| stat |
Applying Machine Learning To Maize Traits Prediction | Heterosis is the improved or increased function of any biological quality in
a hybrid offspring. We have studied yet the largest maize SNP dataset for
traits prediction. We develop linear and non-linear models which consider
relationships between different hybrids as well as other effect. Specially
designed model proved to be efficient and robust in prediction maize's traits.
| 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.