Aaditya Ramdas
Assistant Professor Carnegie Mellon University
- Pittsburgh PA
Aaditya Ramdas' research is aimed at solving basic problems in science and technology.
Biography
Areas of Expertise
Media Appearances
Carnegie Mellon Leads NSF AI Institute for Societal Decision Making
Carnegie Mellon University online
2023-05-04
Leading this work will be Ariel Procaccia, a professor of computer science at Harvard, and Aaditya Ramdas, an assistant professor in CMU’s Department of Statistics & Data Science(opens in new window) and Machine Learning Department.
“When AI or humans predict how a particular situation will evolve or propose varying options to take because of different underlying perceptions of risk and utility, it is important to think about how best to elicit these complex preferences and combine them into a group decision,” Ramdas said. “In a setting where these agents make repeated decisions, we hope to design algorithms that can learn from experience how to combine these decisions — from AI or humans with possibly different individual incentives — toward a common group goal.”
Media
Industry Expertise
Education
Indian Institute of Technology
B.S.
Computer Science and Engineering
Carnegie Mellon University
Ph.D.
Statistics and Machine Learning
Articles
Catoni-style confidence sequences for heavy-tailed mean estimation
Stochastic Processes and Applications2023
A confidence sequence (CS) is a sequence of confidence intervals that is valid at arbitrary data-dependent stopping times. These are useful in applications like A/B testing, multi-armed bandits, off-policy evaluation, election auditing, etc. We present three approaches to constructing a confidence sequence for the population mean, under the minimal assumption that only an upper bound σ2 on the variance is known. While previous works rely on light-tail assumptions like boundedness or subGaussianity (under which all moments of a distribution exist), the confidence sequences in our work are able to handle data from a wide range of heavy-tailed distributions.
Game-theoretic statistics and safe anytime-valid inference
Statistical Science2023
Safe anytime-valid inference (SAVI) provides measures of statistical evidence and certainty -- e-processes for testing and confidence sequences for estimation -- that remain valid at all stopping times, accommodating continuous monitoring and analysis of accumulating data and optional stopping or continuation for any reason. These measures crucially rely on test martingales, which are nonnegative martingales starting at one. Since a test martingale is the wealth process of a player in a betting game, SAVI centrally employs game-theoretic intuition, language and mathematics. We summarize the SAVI goals and philosophy, and report recent advances in testing composite hypotheses and estimating functionals in nonparametric settings.
Estimating means of bounded random variables by betting
Journal of the Royal Statistical Society2023
This paper derives confidence intervals (CI) and time-uniform confidence sequences (CS) for the classical problem of estimating an unknown mean from bounded observations. We present a general approach for deriving concentration bounds, that can be seen as a generalization and improvement of the celebrated Chernoff method. At its heart, it is based on a class of composite nonnegative martingales, with strong connections to testing by betting and the method of mixtures. We show how to extend these ideas to sampling without replacement, another heavily studied problem. In all cases, our bounds are adaptive to the unknown variance, and empirically vastly outperform existing approaches based on Hoeffding or empirical Bernstein inequalities and their recent supermartingale generalizations. In short, we establish a new state-of-the-art for four fundamental problems: CSs and CIs for bounded means, when sampling with and without replacement.
Comparing sequential forecasters
Operations Research2023
Consider two forecasters, each making a single prediction for a sequence of events over time. We ask a relatively basic question: how might we compare these forecasters, either online or post-hoc, while avoiding unverifiable assumptions on how the forecasts and outcomes were generated? In this paper, we present a rigorous answer to this question by designing novel sequential inference procedures for estimating the time-varying difference in forecast scores. To do this, we employ confidence sequences (CS), which are sequences of confidence intervals that can be continuously monitored and are valid at arbitrary data-dependent stopping times ("anytime-valid"). The widths of our CSs are adaptive to the underlying variance of the score differences. Underlying their construction is a game-theoretic statistical framework, in which we further identify e-processes and p-processes for sequentially testing a weak null hypothesis -- whether one forecaster outperforms another on average (rather than always). Our methods do not make distributional assumptions on the forecasts or outcomes; our main theorems apply to any bounded scores, and we later provide alternative methods for unbounded scores. We empirically validate our approaches by comparing real-world baseball and weather forecasters.
E-values as unnormalized weights in multiple testing
Biometrika2023
We study how to combine p-values and e-values, and design multiple testing procedures where both p-values and e-values are available for every hypothesis. Our results provide a new perspective on multiple testing with data-driven weights: while standard weighted multiple testing methods require the weights to deterministically add up to the number of hypotheses being tested, we show that this normalization is not required when the weights are e-values that are independent of the p-values. Such e-values can be obtained in the meta-analysis setting wherein a primary dataset is used to compute p-values, and an independent secondary dataset is used to compute e-values. Going beyond meta-analysis, we showcase settings wherein independent e-values and p-values can be constructed on a single dataset itself. Our procedures can result in a substantial increase in power, especially if the non-null hypotheses have e-values much larger than one.
