A5101961140- Statistical Methods and Inference
- Statistical Methods and Bayesian Inference
- Advanced Causal Inference Techniques
- Bayesian Methods and Mixture Models
- Electoral Systems and Political Participation
- Markov Chains and Monte Carlo Methods
- European and International Law Studies
- Simulation Techniques and Applications
- Scientific Computing and Data Management
- Survey Sampling and Estimation Techniques
- Judicial and Constitutional Studies
- Distributed Sensor Networks and Detection Algorithms
- Privacy-Preserving Technologies in Data
- Gaussian Processes and Bayesian Inference
- Cancer Genomics and Diagnostics
- Advanced Software Engineering Methodologies
- Advanced Text Analysis Techniques
- Parallel Computing and Optimization Techniques
- Bayesian Modeling and Causal Inference
- Philosophy and History of Science
- Service-Oriented Architecture and Web Services
- Ethics in Clinical Research
- Economic Policies and Impacts
- Media Influence and Politics
- Explainable Artificial Intelligence (XAI)
University of Michigan–Ann Arbor
2009-2023
Meta (United States)
2023
Genomics (United Kingdom)
2019
Georgetown University
2017
Berkeley College
2010-2016
University of California, Berkeley
2009-2016
University of California Office of the President
2016
Executive Office of the President
2016
University of Washington
2016
Software (Spain)
2015
We introduce <b>MCMCpack</b>, an R package that contains functions to perform Bayesian inference using posterior simulation for a number of statistical models. In addition code can be used fit commonly models, <b>MCMCpack</b> also some useful utility functions, including additional density and pseudo-random generators distributions, general purpose Metropolis sampling algorithm, tools visualization.
In this paper, we discuss an estimator for average treatment effects (ATEs) known as the augmented inverse propensity weighted (AIPW) estimator. This has attractive theoretical properties and only requires practitioners to do two things they are already comfortable with: (1) specify a binary regression model score, (2) outcome variable. Perhaps most interesting property of is its so-called “double robustness.” Put simply, remains consistent ATE if either score or misspecified but other...
Many situations exist in which a latent construct has both ordinal and continuous indicators. This presents problem for the applied researcher because standard measurement models are not designed to accommodate mixed data. I address this by formulating model that is appropriate such multivariate responses. unifies normal theory factor analysis item response detail Markov chain Monte Carlo algorithm fitting. apply cross-national data on political-economic risk find works well. Software...
Legislative voting records are an important source of information about legislator preferences, intraparty cohesiveness, and the divisiveness various policy issues. Standard methods analyzing a legislative record tend to have serious drawbacks when applied legislatures, such as United Kingdom House Commons, that feature highly disciplined parties, strategic voting, large amounts missing data. We present method (based on Dirichlet process mixture model) for does not suffer from these same...
Case studies appear prominently in political science, sociology, and other social science fields. A scholar employing a case study research design an effort to estimate causal effects must confront the question, how should cases be selected for analysis? This question is important because results derived from program ultimately unavoidably rely on criteria used select cases. While matter of selection at forefront design, analytical framework that can address it comprehensive way has yet...
Meta has traditionally relied on using CPU-based servers for running inference workloads, specifically Deep Learning Recommendation Models (DLRM), but the increasing compute and memory requirements of these models have pushed company towards specialized solutions such as GPUs or other hardware accelerators. This paper describes company's effort in constructing its first silicon designed recommendation systems; it accelerator architecture platform design, software stack enabling optimizing...
Research efforts at network design in the area of Autonomic Networking and Self-Managing Networks have reached a maturity level that forms strong foundation toward standardization architectural principles Future Internet. Therefore, an Industry Specification Group (ISG) on engineering for self-managing Internet (AFI) has been established under auspices European Telecommunications Standards Institute (ETSI). Upon its creation, main stakeholders agreed to harmonize previous developments most...
The entire electricity infrastructure and associated socio-technical system including transmission distribution networks, the operator, suppliers, generators, consumers market mechanisms will need to evolve realize full potential of smart-grids. At heart this evolution is integration information communication technology (ICT) energy infrastructures for increasingly decentralized development, monitoring management a resilient grid. This paper identifies challenges four key areas future...
Summary Despite its potential pitfalls, ecological inference is an unavoidable part of some quantitative settings, including US voting rights litigation. In such applications, the analyst will typically encounter two-way tables with more than two rows and columns. Although several methods are currently available for 2×2 tables, there fewer options analysing general R×C virtually none that model counts as opposed to fractions. We propose a count method respects bounds deterministically,...
In the years since first human genome was sequenced at a cost of over $3 billion, technological advancements have driven price below $1,000, making personal sequencing affordable to many people. Personal has potential enable better disease prevention, more accurate diagnoses, and personalized therapies. Furthermore, sharing genomic data with researchers promises identification causes diseases development new However, costs, privacy concerns, regulatory restrictions, technical challenges...
Despite its shortcomings, cross-level or ecological inference remains a necessary part of some areas quantitative inference, including in United States voting rights litigation. Ecological suffers from lack identification that, most agree, is best addressed by incorporating individual-level data into the model. In this paper we test limits such an incorporation attempting it context drawing inferences about racial patterns using combination exit poll and precinct-level data; accurate...
The <b>Scythe</b> Statistical Library is an open source C++ library for statistical computation. It includes a suite of matrix manipulation functions, pseudo-random number generators, and numerical optimization routines. Programs written using are generally much faster than those in commonly used interpreted languages, such as R proglang{MATLAB}; can be compiled on any system with the GNU GCC compiler (and perhaps other compilers). One primary design goals developers has been ease use...
Our goal in this paper is to provide a formal explanation for how within-unit causal process information (i.e., data on posttreatment variables and partial counterfactuals) can help inform inferences relating total effects—the overall effect of an explanatory variable outcome variable. The basic idea that, many applications, researchers may be able make more plausible assumptions conditional the value than they would do unconditionally. As become available variable, these active about...
Presidents often campaign on behalf of candidates during elections. Do these visits increase the probability that candidate will win? While one might attempt to answer this question by adjusting for observed covariates, such an approach is plagued serious data limitations. In paper we pursue a different approach. Namely, ask: what, if anything, should infer about causal effect presidential visit using simple cross-tabulation data? We take Bayesian problem and show willing use substantive...
Abstract The growing number of health-data breaches, the use genomic databases for law enforcement purposes and lack transparency personal-genomics companies are raising unprecedented privacy concerns. To enable a secure exploration datasets with controlled transparent data access, we propose novel approach that combines cryptographic privacy-preserving technologies, such as homomorphic encryption multi-party computation, auditability blockchains. This provides strong security guarantees...