Beepul Bharti

Hi everyone, my name is Beepul Bharti! I am a PhD student in the Biomedical Engineering Department at Johns Hopkins University, where I am very fortunate to be advised by Dr. Jeremias Sulam and be affiliated with the JHU Mathematical Institute of Data Science (MINDS) . My PhD research focuses on the statistical understanding of machine learning (ML), the interpretability and fairness of ML algorithms, and how to responsibly utilize/deploy these algorithms in high-stakes decision areas such as governance and healthcare.

Prior to my PhD, I completed my BS in Biomedical Engineering, BA in Mathematics, and minor in Chemistry from Duke University.

In my spare time, I enjoy running, playing soccer, watching basketball, and spending time with family and friends.

news

Oct, 2024	Our latest preprint “Sufficient and Necessary Explanations (and What Lies in Between)” has been posted on arxiv!
Aug, 2024	I spent the Summer of 2024 interning at Genentech as a ML/AI Intern working with the Deep Learning Theory and Algorithms team within the Biology Research/AI Development department. Thank you Genentech and specifically Alex Tseng for hosting me!
Dec, 2023	Our paper “SHAP-XRT: The Shapley Value Meets Conditional Independence Testing” has been accepted to Transactions on Machine Learning Research!
Sep, 2023	Our paper “Estimating and Controlling for Equalized Odds via Sensitive Attribute Predictors” has been accepted to NeurIPS 2023!
Sep, 2023	I had a great time attending the SLMath Introductory Workshop: Algorithms, Fairness,and Equity! Thank you to SLMath for hosting us and to all the speakers for their great talks!
Jan, 2022	Awarded the 2022 Spring & Summer Data Science Fellowship by (MINDS)

preprints

publications

Dec. 13, 2023
NeurIPS

Estimating and Controlling for Equalized Odds via Sensitive Attribute Predictors

Bharti, B., Yi, P., and Sulam, J.

Abstract

As the use of machine learning models in real world high-stakes decision settings continues to grow, it is highly important that we are able to audit and control for any potential fairness violations these models may exhibit towards certain groups. To do so, one naturally requires access to sensitive attributes, such as demographics, biological sex, or other potentially sensitive features that determine group membership. Unfortunately, in many settings, this information is often unavailable. In this work we study the well known equalized odds (EOD) definition of fairness. In a setting without sensitive attributes, we first provide tight and computable upper bounds for the EOD violation of a predictor. These bounds precisely reflect the worst possible EOD violation. Second, we demonstrate how one can provably control the worst-case EOD by a new post-processing correction method. Our results characterize when directly controlling for EOD with respect to the predicted sensitive attributes is – and when is not – optimal when it comes to controlling worst-case EOD. Our results hold under assumptions that are milder than previous works, and we illustrate these results with experiments on synthetic and real datasets.
Dec. 12, 2023
TMLR

SHAP-XRT: The Shapley Value Meets Conditional Independence Testing

Teneggi, J.*, Bharti, B.*, Romano, Y., and Sulam, J.

Abstract

The complex nature of artificial neural networks raises concerns on their reliability, trustworthiness, and fairness in real-world scenarios. The Shapley value—a solution concept from game theory—is one of the most popular explanation methods for machine learning models. More traditionally, from a statistical perspective, feature importance is defined in terms of conditional independence. So far, these two approaches to interpretability and feature importance have been considered separate and distinct. In this work, we show that Shapley-based explanation methods and conditional independence testing are closely related. We introduce the SHAPley-EXplanation Randomization Test (SHAP-XRT), a testing procedure inspired by the Conditional Randomization Test (CRT) for a specific notion of local (i.e., on a sample) conditional independence. With it, we prove that for binary classification problems, the marginal contributions in the Shapley value provide lower and upper bounds to the p-values of their respective tests. Furthermore, we show that the Shapley value itself provides an upper bound to the p-value of a global (i.e., overall) null hypothesis. As a result, we further our understanding of Shapley-based explanation methods from a novel perspective and characterize under which conditions one can make statistically valid claims about feature importance via the Shapley value.
Aug. 30, 2022
CAR

Sex imbalance produces biased deep learning models for knee osteoarthritis detection

Li, D., Bharti, B., Wei, J., Sulam, J., and Yi, P.