Generative AI
Differential privacy
High dimensional statistics
Best subset selection
Robust Inference
Distributed learning
Reinforcement learning
Saptarshi Roy, Alessand Rinaldo, Purnamrita Sarkar (2026). Low-Dimensional Adaptation of Rectified Flow: A New Perspective through the Lens of Diffusion and Stochastic Localization.
Vansh Bansal *, Saptarshi Roy *, Purnamrita Sarkar, Alessandro Rinaldo (2025). On the convergence and straightness of Rectified Flow (Accepted in AISTATS 2026)
Subha Maity, Saptarshi Roy, Songkai Xue, Mikhail Yurochkin, Yuekai Sun (2025). How does overparametrization affect performance on minority groups? (Accepted in Transactions on Machine Learning Research)
Saptarshi Roy, Ambuj Tewari, Ziwei Zhu (2025). Understanding Best Subset Selection: A Tale of Two C(omplex)ities (Accepted in Electronic Journal of Statistics)
Sunrit Chakraborty *, Saptarshi Roy *, Debabrota Basu (2024). FLIPHAT: Joint Differential Privacy for High Dimensional Sparse Linear Bandits (Accepted at AISTATS 2025, To appear) (* = Equal contribution)
Saptarshi Roy, Zehua Wang, Ambuj Tewari (2023). On the Computational Complexity of Private High-dimensional Model Selection via Exponential Mechanism (Accepted at Neurips 2024)
Ryan Swope , Amol Khanna , Philip Doldo , Saptarshi Roy , Edward Raff (2024). Feature Selection from Differentially Private Correlations (Accepted at 17th ACM Workshop on Artificial Intelligence and Security)
Saptarshi Roy, Ambuj Tewari, Ziwei Zhu (2024). High-dimensional variable selection with heterogeneous signals: A precise asymptotic perspective (Accepted in Bernoulli)
Sunrit Chakraborty *, Saptarshi Roy *, Ambuj Tewari (2023). Thompson Sampling for High-Dimensional Sparse Linear Contextual Bandits (Accepted in ICML 2023) (* = Equal contribution)
Saptarshi Roy, Kaustav Chakraborty, Somnath Bhadra, Ayanendranath Basu (2019). Density Power Downweighting and Robust Inference: Some New Strategies (Accepted in Journal of Mathematics and Statistics)
Saptarshi Roy (2024). Statistics in the Modern Era: High Dimensions, Decision-Making, and Privacy