Statistics and Data Science Seminar

Abstract:

Many algorithms take prohibitively long to run on modern, large data sets. But even in complex data sets, many data points may be at least partially redundant for some task of interest. So one might instead construct and use a weighted subset of the data (called a coreset) that is much smaller than the original dataset. Typically running algorithms on a much smaller data set will take much less computing time, but it remains to understand whether the output can be widely useful. (1) In particular, can running an analysis on a smaller coreset yield answers close to those from running on the full data set? (2) And can useful coresets be constructed automatically for new analyses, with minimal extra work from the user? We answer in the affirmative for a wide variety of problems in Bayesian inference. We demonstrate how to construct Bayesian coresets as an automatic, practical pre-processing step. We prove that our method provides geometric decay in relevant approximation error as a function of coreset size. Empirical analysis shows that our method reduces approximation error by orders of magnitude relative to uniform random subsampling of data. Though we focus on Bayesian methods here, we also show that our construction can be applied in other domains.

Biography:

Tamara Broderick is an Associate Professor in the Department of Electrical Engineering and Computer Science at MIT. She is a member of the MIT Computer Science and Artificial Intelligence Laboratory (CSAIL), the MIT Statistics and Data Science Center, and the Institute for Data, Systems, and Society (IDSS). She completed her Ph.D. in Statistics at the University of California, Berkeley in 2014. Previously, she received an AB in Mathematics from Princeton University (2007), a Master of Advanced Study for completion of Part III of the Mathematical Tripos from the University of Cambridge (2008), an MPhil by research in Physics from the University of Cambridge (2009), and an MS in Computer Science from the University of California, Berkeley (2013). Her recent research has focused on developing and analyzing models for scalable Bayesian machine learning. She has been awarded an AISTATS Notable Paper Award (2019), NSF CAREER Award (2018), a Sloan Research Fellowship (2018), an Army Research Office Young Investigator Program award (2017), Google Faculty Research Awards, an Amazon Research Award, the ISBA Lifetime Members Junior Researcher Award, the Savage Award (for an outstanding doctoral dissertation in Bayesian theory and methods), the Evelyn Fix Memorial Medal and Citation (for the Ph.D. student on the Berkeley campus showing the greatest promise in statistical research), the Berkeley Fellowship, an NSF Graduate Research Fellowship, a Marshall Scholarship, and the Phi Beta Kappa Prize (for the graduating Princeton senior with the highest academic average).

–

The MIT Statistics and Data Science Center hosts guest lecturers from around the world in this weekly seminar.

Abstract:
Over 500K scientific articles have been published since 1999 with the word network in the title. And the vast majority of these report network summary statistics of one type or another. However, these numbers are rarely accompanied by any quantification of uncertainty. Yet any error inherent in the measurements underlying the construction of the network, or in the network construction procedure itself, necessarily must propagate to any summary statistics reported. Perhaps surprisingly, there is little in the way of formal statistical methodology for this problem. I summarize results from our recent work, for the case of estimating the density of low-order subgraphs. Under a simple model of network error, we show that consistent estimation of such densities is impossible when the rates of error are unknown and only a single network is observed. We then develop method-of-moment estimators of subgraph density and error rates for the case where a minimal number of network replicates are available (i.e., just 2 or 3). These estimators are shown to be asymptotically normal as the number of vertices increases to infinity. We also provide confidence intervals for quantifying the uncertainty in these estimates, implemented through a novel bootstrap algorithm. We illustrate the use of our estimators in the context of gene coexpression networks — the correction for measurement error is found to have potentially substantial impact on standard summary statistics. This is joint work with Qiwei Yao and Jinyuan Chang.

Biography:
Eric Kolaczyk is a Professor of Statistics and Director of the Program in Statistics in the Department of Mathematics & Statistics at Boston University. He is also a university Data Science Faculty Fellow, and affiliated with the Division of Systems Engineering and the Programs in Bioinformatics and in Computational Neuroscience. His current research interests revolve mainly around the statistical analysis of network-indexed data, including both theory/methods development and collaborative research. He has published several books on the topic of network analysis. As an associate editor, he has served on the boards of JASA and JRSS-B in statistics, IEEE IP and TNSE in engineering, and SIMODS in mathematics. Currently he is the co-chair of the NAS Roundtable on Data Science Education. He is an elected fellow of the AAAS, ASA, and IMS, an elected senior member of the IEEE, and an elected member of the ISI.

MIT Statistics and Data Science Center host guest lecturers from around the world in this weekly seminar.

Abstract: Despite tremendous methodological advances in causal inference, there remain significant gaps in our understanding of the fundamental statistical limits of estimating various causal estimands from observational data. In this talk I will survey some recent work that aims to make some progress towards closing these gaps. Particularly, I will discuss the fundamental limits of estimating various important causal estimands under classical smoothness assumptions, under new “structure-agnostic” assumptions, in a discrete setup, and under partial smoothness assumptions. Via these fundamental limits we will also attempt to understand the optimality/sub-optimality of simple, practical procedures for estimating causal effects from observational data.

The talk will be based on joint works with Yanjun Han, Edward Kennedy, James Robins, Larry Wasserman and Tiger Zeng.

Bio: Sivaraman is a Professor with a joint appointment in the Department of Statistics and Data Science, and the Machine Learning Department at Carnegie Mellon. This year he is a long-term visitor at the Simons Institute, and a visiting scholar in the Department of Statistics at Berkeley. Prior to Carnegie Mellon he was a postdoctoral researcher at UC Berkeley working with Martin Wainwright and Bin Yu. His research interests are broadly in statistical machine learning and algorithmic statistics. Particular areas that he is currently most fascinated by include optimal transport, causal inference, robust statistics and minimax hypothesis testing.

Abstract: As ML models become increasingly powerful, it is an attractive proposition to use them in important decision making pipelines, in collaboration with human decision makers. But how should a human being and a machine learning model collaborate to reach decisions that are better than either of them could achieve on their own? If the human and the ML model were perfect Bayesians, operating in a setting with a commonly known and correctly specified prior, Aumann’s classical agreement theorem would give us one answer: they could engage in conversation about the task at hand, and their conversation would be guaranteed to converge to (accuracy-improving) agreement. This classical result however would require making many implausible assumptions, both about the knowledge and computational power of both parties. We show how to recover similar (and more general) results using only computationally and statistically tractable assumptions, which substantially relax full Bayesian rationality. We further give weak-learning conditions under which this collaboration will result in “information aggregation” — i.e. predictions that are as accurate as could have been made by a model that had access to -both- party’s observations, even though neither party in the interaction actually has access to these pooled observations.

Joint work with Natalie Collina, Varun Gupta, and Surbhi Goel, based on a paper that will appear in STOC 2025, and with Natalie Collina, Ira Globus-Harris, Varun Gupta, Surbhi Goel, and Mirah Shi based on a new preprint.

Bio: Aaron Roth is the Henry Salvatori Professor of Computer and Cognitive Science, in the Computer and Information Sciences department at the University of Pennsylvania, with a secondary appointment in the Wharton statistics department. He is affiliated with the Warren Center for Network and Data Science, and co-director of the Networked and Social Systems Engineering (NETS) program.  He is also an Amazon Scholar at Amazon AWS. He is the recipient of the Hans Sigrist Prize, a Presidential Early Career Award for Scientists and Engineers (PECASE), an Alfred P. Sloan Research Fellowship, an NSF CAREER award, and research awards from Yahoo, Amazon, and Google.  His research focuses on the algorithmic foundations of data privacy, algorithmic fairness, game theory, learning theory, and machine learning.  Together with Cynthia Dwork, he is the author of the book “The Algorithmic Foundations of Differential Privacy.” Together with Michael Kearns, he is the author of “The Ethical Algorithm”.

Abstract: In the context of linear regression, we construct a data-driven convex loss function with respect to which empirical risk minimisation yields optimal asymptotic variance in the downstream estimation of the regression coefficients. Our semiparametric approach targets the best decreasing approximation of the derivative of the log-density of the noise distribution. At the population level, this fitting process is a nonparametric extension of score matching, corresponding to a log-concave projection of the noise distribution with respect to the Fisher divergence. The procedure is computationally efficient, and we prove that our procedure attains the minimal asymptotic covariance among all convex M-estimators. As an example of a non-log-concave setting, for Cauchy errors, the optimal convex loss function is Huber-like, and our procedure yields an asymptotic efficiency greater than 0.87 relative to the oracle maximum likelihood estimator of the regression coefficients that uses knowledge of this error distribution; in this sense, we obtain robustness without sacrificing much efficiency.

Bio: Richard Samworth obtained his PhD in Statistics from the University of Cambridge in 2004, and has remained in Cambridge since, becoming a full professor in 2013 and the Professor of Statistical Science in 2017.  His main research interests are in high-dimensional and nonparametric statistics; he has developed methods and theory for shape-constrained inference, missing data, subgroup selection, data perturbation techniques (subsampling, the bootstrap, random projections, knockoffs), changepoint estimation and independence testing, amongst others. Richard currently holds a European Research Council Advanced Grant.  He received the COPSS Presidents’ Award in 2018, was elected a Fellow of the Royal Society in 2021 and served as co-editor of the Annals of Statistics (2019-2021).

Abstract: One dominant approach to evaluate the causal effect of a treatment is through panel data analysis, whereby the behaviors of multiple units are observed over time. The information across time and units motivates two general approaches: (i) horizontal regression (i.e., unconfoundedness), which exploits time series patterns, and (ii) vertical regression (e.g., synthetic controls), which exploits cross-sectional patterns. Conventional wisdom often considers the two approaches to be different. We establish this position to be partly false for estimation but generally true for inference. In the absence of any assumptions, we show that both approaches yield algebraically equivalent point estimates for several standard estimators. However, the source of randomness assumed by each approach leads to a distinct estimand and quantification of uncertainty even for the same point estimate. This emphasizes that researchers should carefully consider where the randomness stems from in their data, as it has direct implications for the accuracy of inference.

Bio:
Dennis Shen is an assistant professor in the Data Sciences and Operations Department at the USC Marshall School of Business. Before joining USC, he was a FODSI postdoctoral fellow at the Simons Institute at UC Berkeley. He also served as a technical consultant for Uber Technologies and TauRx Therapeutics. He has received several recognition for his work, including the INFORMS George B. Dantzig Dissertation Award (2nd to his esteemed colleague, Somya) and MIT George Sprowls PhD Thesis Award in Artificial Intelligence & Decision-making.

Abstract:

(with Iman Modarressi and Amar Venugopal; arxiv.org/abs/2503.00725

We propose a machine-learning tool that yields causal inference on text in randomized trials. Based on a simple econometric framework in which text may capture outcomes of interest, our procedure addresses three questions: First, is the text affected by the treatment? Second, which outcomes is the effect on? And third, how complete is our description of causal effects? To answer all three questions, our approach uses large language models (LLMs) that suggest systematic differences across two groups of text documents and then provides valid inference based on costly validation. Specifically, we highlight the need for sample splitting to allow for statistical validation of LLM outputs, as well as the need for human labeling to validate substantive claims about how documents differ across groups. We illustrate the tool in a proof-of-concept application using abstracts of academic manuscripts.

Bio:
Jann is an econometrician in the OIT group at Stanford GSB. His current research focuses broadly on two related themes: (1) high-dimensional and robust causal inference, including work on using machine learning to improve inferences from randomized trials, robust inference in panel data, synthetic control, matching estimation, highly over-parametrized models, and high-dimensional outcome data; and (2) data-driven decision-making with misaligned objectives, including work on algorithmic fairness, human–AI interaction, the regulation of algorithms, and the design of pre-analysis plans. He holds a PhD in economics from Harvard University.