Neyman Seminar

The Neyman seminar is the statistics seminar in the Department. Historically, it has been focused on applications of Statistics to other fields. Nowadays, it has a very broad scope, with topics ranging from applications of statistics to theory.

The seminar is held on Wednesdays from 4 to 5 in the Jerzy Neyman room, 1011 Evans.

Details of individual seminar events are published in the campus' event system.

You can sign up to the department's seminars@stat mailing list to receive related announcements.

Add this series of events to your calendar: ICAL or XML

Recent & Upcoming Neyman Seminars

Rerandomization and Regression Adjustment
Peng Ding, UC Berkeley
Aug 22, 2018 4:00pm
1011 Evans Hall
Abstract:
Randomization is a basis for the statistical inference of treatment effects without assumptions on the outcome generating process. Appropriately using covariates further yields more precise estimators in randomized experiments. In his seminal work Design of Experiments, R. A. Fisher suggested blocking on discrete covariates in the design stage and conducting the analysis of covariance (ANCOVA) in...
Likelihood Ratio Test for Stochastic Block Models with Bounded Degrees
Yang Feng, Columbia University
Aug 29, 2018 4:00pm
1011 Evans Hall
Abstract:
A fundamental problem in network data analysis is to test whether a network contains statistical significant communities. We study this problem in the stochastic block model context by testing H0: Erdos-Renyi model vs. H1: stochastic block model. This problem serves as the foundation for many other problems including the testing-based methods for determining the number of communities and...
Stochastic Gradient Descent: Strong convergence guarantees -- without parameter tuning
Rachel Ward, UT Austin
Aug 30, 2018 4:00pm
60 Evans Hall
Abstract:
Stochastic Gradient Descent is the basic optimization algorithm behind powerful deep learning architectures which are becoming increasingly omnipresent in society. However, existing theoretical guarantees of convergence rely on knowing certain properties of the optimization problem such as maximal curvature and noise level which are not known a priori in practice. Thus, in practice, hyper...
AdaPT: An interactive procedure for multiple testing with side information
Will Fithian, UC Berkeley
Sep 5, 2018 4:00pm
1011 Evans Hall
Abstract:
We consider the problem of multiple hypothesis testing with generic side information: for each hypothesis we observe both a p-value and some predictor encoding contextual information about the hypothesis. For large-scale problems, adaptively focusing power on the more promising hypotheses (those more likely to yield discoveries) can lead to much more powerful multiple testing procedures. We...
How to estimate the mean of a random vector?
Gabor Lugosi, Pompeu Fabra University
Sep 12, 2018 4:00pm
1011 Evans Hall
Abstract:
Given n independent, identically distributed copies of a random vector, one is interested in estimating the expected value. Perhaps surprisingly, there are still open questions concerning this very basic problem in statistics. The goal is to construct estimators that are close to the true mean with high probability, with respect to some given norm. In this talk we are primarily...
Correcting Bias in Eigenvectors of Financial Covariance Matrices
Alex Papanicolaou, UC Berkeley
Sep 19, 2018 4:00pm
1011 Evans Hall
Abstract:
There is a source of bias in the sample eigenvectors of financial covariance matrices, when unchecked, distorts weights of minimum variance portfolios and leads to risk forecasts that are severely biased downward. Recent work with Lisa Goldberg and Alex Shkolnik develops an eigenvector bias correction. Our approach is distinct from the regularization and eigenvalue shrinkage methods found in the...
Unraveling Controversy on Vexed Environmental Risks
Jeroen P. van der Sluijs, University of Bergen and Utrecht University
Sep 26, 2018 4:00pm
1011 Evans Hall
Abstract:
Scientific assessment of many contemporary risks is plagued by controversy, persistent uncertainty, and polarized societal contexts. Decision makers often become mired in contested evidence, beset by uncertainties and contradictions. This leads to inaction on early warnings, paralysis-by-analysis, and erodes trust in science and its institutions. But why do controversies persist? A new...
Statistical challenges in casualty estimation
Kristian Lum, Human Rights Data Analysis Group
Oct 3, 2018 4:00pm
1011 Evans Hall
Abstract:
An accurate understanding of the magnitude and dynamics of casualties during a conflict is important for a variety of reasons, including historical memory, retrospective policy analysis, and assigning culpability for human rights violations. However, during times of conflict and their aftermath, collecting a complete or representative sample of casualties can be difficult if not impossible. One...
To persist or not to persist?
Sebastian Schreiber, UC Davis
Oct 10, 2018 4:00pm
1011 Evans Hall
Abstract:
Two long standing, fundamental questions in biology are "Under what conditions do populations persist or go extinct? When do interacting species coexist?" The answers to these questions are essential for guiding conservation efforts and identifying mechanisms that maintain biodiversity. Mathematical models play an important role in identifying these mechanisms and, when coupled with empirical...
Learning in Google Ads, Machines and People
Niall Cardin, Google
Oct 17, 2018 4:00pm
1011 Evans Hall
Abstract:
This talk is in two parts, both of which discuss interesting uses of experiments in Google search ads. In part 1 I discuss how we can inject randomness into our system to get causal inference in a machine learning setting. In part 2. I talk about experiment designs to measure how users learn in response to ads on Google.com.
Safe Learning in Robotics
Claire Tomlin, UC Berkeley
Oct 24, 2018 4:00pm
1011 Evans Hall
Abstract:
A great deal of research in recent years has focused on robot learning. In many applications, guarantees that specifications are satisfied throughout the learning process are paramount. For the safety specification, we present a controller synthesis technique based on the computation of reachable sets, using optimal control and game theory. In the first part of the talk, we will review these...
Why Deep Learning Works: Traditional and Heavy-Tailed Implicit Self-Regularization in Deep Neural Networks
Michael W. Mahoney, UC Berkeley
Nov 7, 2018 4:00pm
1011 Evans Hall
Abstract:
Random Matrix Theory (RMT) is applied to analyze the weight matrices of Deep Neural Networks (DNNs), including both production quality, pre-trained models and smaller models trained from scratch. Empirical and theoretical results clearly indicate that the DNN training process itself implicitly implements a form of self-regularization, implicitly sculpting a more regularized energy or penalty...
See more events »