About me
I am a Ph.D. student in Electrical and Computer Engineering at the University of Illinois, Urbana-Champaign, supervised by Prof. R. Srikant.
I graduated with a dual degree (Bachelor’s and Master’s) from the Indian Institute of Technology, Kharagpur, majoring in Electronics and Electrical Communication Engineering and minoring in Computer Science in 2021.
I spent the Fall of 2023 working at Google Research on an “AI for Social Good” project, where I studied the use of data-driven approaches to efficiently allocate scarce resources and encourage patients to seek healthcare in rural India. I worked under the supervision of Prof. Milind Tambe and collaborated closely with his team at Harvard University. The project involved designing sequential decision-making policies using the Restless Multi-armed Bandit framework.
Here is my curriculum vitae.
Research Interests
I have broad interests in the theoretical understanding of machine learning. Currently, I am exploring distributionally robust reinforcement learning in the average cost framework. Previously I worked on knowledge distillation from a theoretical perspective, particularly using the techniques of the neural tangent kernel. I also investigated Type dependent estimate for Crowdsourcing. Additionally, I am interested in sequential decision-making problems, including restless bandits, slicing in modern communication networks, scheduling in sensor networks, etc.
Updates
2025
- The Crowdsourcing work is accepted for a talk at INFORMS Applied Probability Society Conference to be held in July.
- Presented the Knowledge Distillation work at Learning for Dynamics and Control (L4DC) 2025.
- Paper: GUOJUN XIONG, Haichuan Wang, Yuqi Pan, Saptarshi Mandal, Sanket Shah, Niclas Boehmer, Milind Tambe. Finite-Horizon Single-Pull Restless Bandits: An Efficient Index Policy For Scarce Resource Allocation, has been accepted for publication as a full paper at International Conference on Autonomous Agents and Multiagent Systems (AAMAS) 2025.
2024
- Paper: Saptarshi Mandal, Xiaojun Lin and R. Srikant. A Theoretical Analysis of Soft-Label vs Hard-Label Training in Neural Networks, has been accepted for publication at Learning for Dynamics and Control (L4DC) 2025.
2023
- Paper: Saptarshi Mandal, Seo Taek Kong, Dimitris Katselis, R. Srikant. Spectral Clustering for Crowdsourcing with Inherently Distinct Task Types on arXiv. Its currently under review at Transaction on Machine Learning Research (TMLR).
- Interned at Google Research, AI for Social Good in Fall 2023.
2022
- Passed my Ph.D. qualification exam.
2021
- Joined University of Illinois, Urbana-Champaign as a Ph.D. student in ECE.
- Graduated from Indian Institute of Technology, Kharagpur with a B.Tech. and M.Tech. (major in ECE, minor in CS).
2020
- Interned at Indian Institute of Science in Summer 2020 under the Summer Research Fellow program by the Indian Academy of Science.
2019
- Interned at NVIDIA, Bangalore in Summer 2019.
2016
- Joined Indian Institute of Technology, Kharagpur as a B.Tech. student in ECE.
- Received the KVPY fellowship in India.