Research
My research focuses on provably efficient and resource-aware data-driven decision-making in sequential learning problems arising in Reinforcement Learning, Federated and Distributed Learning, Privacy-Preserving Machine Learning, and Stochastic Optimization. My research agenda broadly has two main thrusts:
- Establish lower bound on feasible performance and characterize the trade-off between learning efficiency and practical and systemic constraints;
- Design provably optimal algorithms for real-world applications that offer practical qualitative benefits such as adaptivity and interpretability.
Recently, I have also started working on developing methods for statistically and computationally efficient inference and alignment of large-scale models. My research employs tools from high-dimensional statistics, probability theory, large-scale optimization, information theory and machine learning and offers a unique perspective into decision making that is based on a blend of information-theoretic, statistical, and systemic design aspects .
You can find my detailed Research Statement here.
Publications and Preprints
Characterizing the Accuracy-Communication-Privacy Trade-off in Distributed Stochastic Convex Optimization
Sudeep Salgia, Nikola Pavlovic, Yuejie Chi, Qing Zhao
International Conference on Artificial Intelligence and Statistics (AISTATS), 2025
Order-Optimal Regret in Distributed Kernel Bandits using Uniform Sampling with Shared Randomness
Nikola Pavlovic, Sudeep Salgia, Qing Zhao
Preliminary version in NeurIPS BDU Workshop, 2024
International Conference on Artificial Intelligence and Statistics (AISTATS), 2025
Differentially Private Kernelized Contextual Bandits
Nikola Pavlovic, Sudeep Salgia, Qing Zhao
International Conference on Artificial Intelligence and Statistics (AISTATS), 2025
Distributed Policy Optimization under Partial Observability: Tractability, Linear Speedup, and Communication Efficiency
Tonghe Zhang, Sudeep Salgia, Yuejie Chi
International Conference on Artificial Intelligence and Statistics (AISTATS), 2025
The Sample-Communication Complexity Trade-off in Federated Q-Learning
Sudeep Salgia, Yuejie Chi
Neural Information Processing Systems (NeurIPS), 2024. Accepted as an oral (top 0.4% of accepted papers)
Random Exploration in Bayesian Optimization: Order-Optimal Regret and Computational Efficiency
Sudeep Salgia, Sattar Vakili, Qing Zhao
International Conference on Machine Learning (ICML), 2024. Resolves an open COLT problem
Adaptive Binning Coincidence Test for Uniformity Testing
Sudeep Salgia, Xinyi Wang, Qing Zhao, Lang Tong
IEEE Transactions on Signal Processing, 2024
A Communication-Efficient Adaptive Algorithm for Federated Learning under Cumulative Regret
Sudeep Salgia, Qing Zhao, Tamir Gabay, Kobi Cohen
IEEE Transactions on Signal Processing, 2024
Collaborative Learning in Kernel-based Bandits
Sudeep Salgia, Sattar Vakili, Qing Zhao
IEEE Transactions on Signal Processing, 2023
Distributed Linear Bandits under Communication Constraints
Sudeep Salgia, Qing Zhao
International Conference on Machine Learning (ICML), 2023
Provably and Practically Efficient Neural Contextual Bandits
Sudeep Salgia, Sattar Vakili, Qing Zhao
International Conference on Machine Learning (ICML), 2023
A perspective on data sharing in digital food safety systems
Chenhao Qian, Yuhan Liu, Cecil Barnett-Neefs, Sudeep Salgia, Omer Serbetci, Aaron Adalja, Jayadev Acharya, Qing Zhao, Renata Ivanek, Martin Wiedmann
Critical Reviews in Food Science and Nutrition
Disagreement-based Active Learning in Online Settings
Boshuang Huang, Sudeep Salgia, Qing Zhao
IEEE Transactions on Signal Processing, 2022
A Domain-Shrinking based Bayesian Optimization Algorithm with Order-Optimal Regret Performance
Sudeep Salgia, Sattar Vakili, Qing Zhao
Neural Information Processing Systems (NeurIPS), 2021
An order-optimal adaptive test plan for noisy group testing under unknown noise models
Sudeep Salgia, Qing Zhao
International Conference on Acoustics, Speech, and Signal Processing (ICASSP), 2021
Stochastic Coordinate Minimization with Progressive Precision for Stochastic Convex Optimization
Sudeep Salgia, Qing Zhao, Sattar Vakili
International Conference on Machine Learning (ICML), 2020
Stochastic Gradient Descent on a Tree: an Adaptive and Robust Approach to Stochastic Convex Optimization
Sattar Vakili, Sudeep Salgia, Qing Zhao
Allerton 2019
On Bandlimited Spatiotemporal Field Sampling with Location and Time Unaware Mobile Sensors
Sudeep Salgia, Animesh Kumar
International Conference on Acoustics, Speech, and Signal Processing (ICASSP), 2018
Other Projects
PhD Thesis: Stochastic Optimization and Learning: An Adaptive and Resource-Efficient Approach
Bandlimited Spatiotemporal Field Sampling with Location and Time Unaware Mobile Senors
UG thesis under the guidance of Prof. Animesh Kumar