Saptarshi Saha is a Ph.D. candidate at the Indian Statistical Institute, Kolkata. His doctoral research is focused on integrating causality into deep learning frameworks to enhance their utility. Beyond his immediate thesis goals, Saptarshi envisions a broader research trajectory aimed at utilizing deep learning for a deeper understanding of cause-and-effect relationships. His aim is to address various challenges such as improving the robustness, explainability, and interpretability of models, addressing issues with limited control over generative models, enhancing generalization performance under varying data distributions, dealing with learning using limited labelled data, promoting fairness in decision-making systems, and more. Saptarshi’s scholarly contributions extend to renowned journals such as TMLR and prominent conferences like ICLR. He has showcased his work at various research fora, such as Amazon Research Day 2023 and the Machine Learning Summer School in Okinawa, 2024.
Saptarshi holds a BS-MS dual degree in mathematics from IISER Kolkata. Throughout his BS-MS studies (2015–2020), he was a recipient of the INSPIRE fellowship from DST, Government of India.
As a Fulbright-Nehru Doctoral Research fellow at the University of Buffalo, Buffalo, NY, Saptarshi is trying to utilize causal knowledge and principles to assess data quality and make informed decisions (in the context of learning with not enough data) regarding samples that need to be labelled (from the large unlabelled dataset) rather than selecting them randomly. He is primarily working on the challenge of efficiently selecting the most relevant samples for labelling while considering budget constraints. This challenge holds excellent relevance not only in academic research but also within the AI industry. Saptarshi is an avid nature photographer and finds solace in the wilderness. His interests extend to culinary adventures, globetrotting, and engaging with diverse cultures. His leisure activities also include playing football and cricket.