Shubhanshu Shekhar

Shubhanshu Shekhar

Postdoctoral Researcher
Department of Statistics and Data Science
Carnegie Mellon University

Email / CV / Github / Google Scholar / Blog-Posts

About

I am a postdoctoral researcher in the Department of Statistics and Data Science, at Carnegie Mellon University working with Prof. Aaditya Ramdas. I obtained my PhD in Electrical Engineering from the University of California, San Diego, where I was advised by Prof. Tara Javidi.

Research

I work on developing principled statistical and algorithmic methods for solving practical problems in machine learning and data science. In particular, I have worked on the following topics:

Adaptive Decision making problems: Bayesian Optimization, Active Learning and Reinforcement Learning.
Resource Allocation problems: learning probability distributions, learning MDPs, fair ML.
Nonparametric statistical inference: permutation-free tests, sequential tests and changepoint detection, uncertainty quantification.

News

Feb. 23 : Uploaded a preprint on sequential changepoint detection.
Dec. 22 : Gave a talk at the IISA Conference 2022, at IISc Bangalore. [Slides]
Dec. 22 : Uploaded a preprint on permutation free independence testing.

Working Papers

Shubhanshu Shekhar, Ilmun Kim and Aaditya Ramdas.
A Permutation-free Kernel Independence Test.

Shubhanshu Shekhar, and Aaditya Ramdas.
Sequential Change Detection via Backward Confidence Sequences.

Shubhanshu Shekhar, Neil Xu and Aaditya Ramdas.
Confidence Sequences for Weighted Sampling without Replacement.

Selected Publications

For the complete list of publications and collaborators, please see my Google Scholar page.

Shubhanshu Shekhar, Ilmun Kim and Aaditya Ramdas.
A Permutation-free Kernel Two-Sample Test
Neural Information Processing Systems (NeurIPS), 2022
Accepted for oral presentation.

Shubhanshu Shekhar and Tara Javidi.
Instance Dependent Regret Analysis of Kernelized Bandits
International Conference on Machine Learning (ICML) 2022

Shubhanshu Shekhar and Aaditya Ramdas.
Nonparametric Two-Sample Testing by Betting
preprint, 2021. (under review at IEEE Transactions on Information Theory)

Shubhanshu Shekhar, Greg Fields, Mohammad Ghavamzadeh and Tara Javidi.
Adaptive Sampling for Minimax Fair Classification
Neural Information Processing Systems (NeurIPS), 2021

Shubhanshu Shekhar, Mohammad Ghavamzadeh and Tara Javidi.
Active Learning for Classification with Abstention
IEEE Journal on Selected Topics in Information Theory, 2021
Among six finalists for Jack K. Wolf Student paper award, ISIT 2020.

Shubhanshu Shekhar, Tara Javidi and Mohammad Ghavamzadeh.
Adaptive Sampling for Learning Probability Distributions.
International Conference on Machine Learning (ICML), 2020

Shubhanshu Shekhar and Tara Javidi.
Gaussian Process Bandits with Adaptive Discretization.
Electronic Journal of Statistics, 2018

Teaching

I worked as a teaching assistant for the following courses.

ECE 101, Linear Systems Fundamentals
ECE 153, Probability and Stochastic Processes for Engineers
ECE 157, Data Networks
ECE 257, Graduate Communication Networks
ECE 267, Graph/Network Algorithms
ECE 272, Dynamical Systems