Parthe Pandit is the Thakur Family Chair Assistant Professor at IIT Bombay with the Center for Machine Intelligence and Data Science. He was an HDSI-Simons Postdoctoral Fellow at the Halıcıoğlu Data Science Institute, UC San Diego. He received his Ph.D. in ECE, and M.S. in Statistics in 2021 both from UCLA, and obtained his undergraduate degree in EE from IIT Bombay. His research focuses on understanding mathematical foundations of machine learning in order to build stable, transparent, high performance, and scalable systems. He received the Jack K. Wolf Student Paper award at IEEE ISIT 2019 for his work on generative models for solving inverse problems.

AI2050 Project

Deep Neural Networks (DNNs) dominate today’s AI revolution, yet their mathematical foundations remain poorly understood. Their high resource demands, lack of interpretability, and reliance on domain-specific heuristics highlight fundamental limitations. In contrast, Kernel Machines (KMs), which led AI prior to 2012, offer a mathematically principled framework characterized by stability, transparency, and simplicity. This project seeks to revitalize KMs by integrating modern engineering advancements—designing high-quality kernel functions and scalable training algorithms—to establish KMs as a sustainable and competitive alternative to DNNs in future AI systems.