Home
Prashant Khanduri
Assistant Professor
Department of Computer Science
Office: 14200.7, 5057 Woodward Ave., Detroit, MI, 48202
About: I am an Assistant Professor in the Department of Computer Science (CS) at Wayne State University (WSU). Previously, I was a Postdoctoral Associate in the Department of Electrical & Computer Engineering (ECE), University of Minnesota (UMN) jointly advised by Prof. Mingyi Hong and Prof. Jia (Kevin) Liu (Assistant Professor of Electrical & Computer Engineering (ECE), The Ohio State University (OSU)). Prior to joining UMN, I was a member of the Sensor Fusion Lab headed by Prof. Pramod K. Varshney. I defended my Ph.D. thesis in 2019 from the Department of Electrical Engineering & Computer Science (EECS), Syracuse University.
I am looking for self-motivated Ph.D. students with an interest in optimization. Interested candidates with a strong background in mathematics, signal processing, statistics, or computer science are encouraged to contact me via email at khanduri.prashant [at] wayne.edu with a copy of their CV and transcripts.
Research Interests: Optimization and Machine Learning, Federated Learning, Robust Optimization, Statistical Learning, Statistical Signal Processing, and Information Theory.
Recent News
[Aug 2023] Our survey paper "An Introduction to Bi-Level Optimization: Foundations and Applications in Signal Processing and Machine Learning," is now available on arXiv.
[July 2023] Two papers accepted by Asilomar, 2023.
"FedAvg for Minimizing Polyak-Lojasiewicz Objectives: The Interpolation Regime," Asilomar, 2023.
"Stochastic Perturbation Based Smoothing for Linearly Constrained Bilevel Optimization," Asilomar (Invited), 2023.
[June 2023] One paper accepted by AdvML-Frontiers, ICML Workshop, 2023., and one by MICCAI, 2023.
"FocalUNETR: A Focal Transformer for Boundary-aware Segmentation of CT Images," MICCAI, 2023.
"Proximal Compositional Optimization for Distributionally Robust Learning," AdvML-Frontiers, ICML Workshop, 2023.
[June 2023] Delivered an invited talk on "Linearly Constrained Bilevel Optimization: An Implicit Gradient Approach," at the SIAM Conference on Optimization (OP23) in Seattle, WA.
[April 2023] Three papers including one joint first-author paper accepted by ICML, 2023.
"Linearly Constrained Bilevel Optimization: A Smoothed Implicit Gradient Approach," ICML, 2023.
"FedAvg Converges to Zero Training Loss Linearly for Overparameterized Multi-Layer Neural Networks," ICML, 2023.
"Prometheus: Taming Sample and Communication Complexities in Constrained Decentralized Stochastic Bilevel Learning," ICML, 2023.
[February 2023] Paper titled "An Implicit Gradient Method for Constrained Bilevel Problems using Barrier Approximation," accepted by ICASSP, 2023.
[January 2023] Our work "Fairness-aware Vision Transformer via Debiased Self-Attention," is now available on arXiv.
[December 2022] Paper titled "DIAMOND: Taming Sample and Communication Complexities in Decentralized Bilevel Optimization," accepted by IEEE INFOCOM, 2023.
[November 2022] Paper titled "With a Little Help from My Friend: Server-Aided Federated Learning with Partial Client Participation," accepted by FL-NeurIPS, 2022.
[October 2022] Delivered an invited talk on "Unconstrained and Constrained Bilevel Optimization: An Implicit Gradient Approach" at the Informs Annual Meeting in Indianapolis, IN.
[September 2022] Delivered an invited talk on "Bilevel Optimization: Algorithms and Guarantees" at the Great Lakes Section of SIAM (GLSIAM) annual meeting hosted by Wayne State University, MI.
[August 2022] Joined the Department of Computer Science (CS) at Wayne State University as Tenure Track Assistant Professor.
[July 2022] Paper titled "INTERACT: Achieving Low Sample and Communication Complexities in Decentralized Bilevel Learning over Networks," accepted by MobiHoc, 2022.
[May 2022] Two papers accepted by ICML, 2022.
"Anarchic Federated Learning," ICML, 2022. (Long Presentation)
"Revisiting and Advancing Fast Adversarial Training Through the Lens of Bi-Level Optimization," ICML, 2022.
[January 2022] One first-author paper accepted by ICLR, 2022, and one paper accepted by ICASSP, 2022.
[November 2021] Paper titled "Byzantine Resilient Non-Convex SCSG with Distributed Batch Gradient Computations," accepted by IEEE Trans. Signal Inf. Process. Netw., 2021.
[October 2021] Two first-author papers accepted by NeurIPS, 2021.