Abstract- People routinely reveal their preferences online, e.g., when choosing search results, videos, or products. Such data is used by algorithms to learn human tastes. Recently, curated datasets of human preferences have been used to fine-tune language models, substantially improving their alignment with human intent. These successes raise a natural question: can recommender systems learn more effectively from comparisons rather than ratings? The talk will trace a path from basic models of choice behaviour to new frameworks for recommender systems. The main focus will be on our theoretical result showing that personalised recommendations can be learned efficiently from comparison data, despite the underlying optimisation problem being nonconvex. I will then describe a bandit formulation that addresses the classical exploration-exploitation trade-off in a novel way. Finally, I’ll share empirical insights motivating richer models of human choice. I will conclude by arguing that learning from human preferences is key to building interactive AI systems that reliably serve human needs.

Bio- Suryanarayana Sankagiri is a postdoctoral researcher at EPFL, Switzerland, in the Information and Network Dynamics Laboratory. He obtained his B.Tech. in Electrical Engineering from IIT Bombay in 2016, and his M.S. and Ph.D. in Electrical and Computer Engineering from the University of Illinois Urbana-Champaign in 2018 and 2022. During his Ph.D., he worked on the design and analysis of network protocols—ranging from message-passing algorithms for graph clustering, to consensus mechanisms in blockchains, to pricing protocols for payment channel networks. At EPFL, his research focuses on recommender systems that learn from comparisons and choices, designing new algorithms as well as models that better capture human behaviour. More broadly, his interests lie in understanding modern human-in-the-loop intelligent systems.

Here is a zoom link for online aattendees: https://zoom.us/j/92550826231?pwd=w0KQ9NmqB3DAHxOnlWkYNXUchUbvzA.1