Optimizing Audio Recommendations for the Long-Term
Participate
Information Systems and Operations Management (ISOM) Department
Speaker : Lucas MAYSTRE, Research scientist at SPOTIFY
in Room Bernard Ramanantsoa
Abstract
"Recommender systems are an essential feature of online streaming services. Most often, the product goal is to enable satisfying recurring user interactions, and is best expressed in terms of long-term user satisfaction. In practice, however, algorithms that power these recommender systems optimize narrow, short-term metrics such as click-through rate or session length. In this talk, I will present recent work from Spotify that attempts to bridge that gap."
Bio
Lucas Maystre is a research scientist at Spotify, working on improving users' long-term engagement and satisfaction. His research interests revolve around probabilistic modeling, causal inference and reinforcement learning. He received a PhD from EPFL, supported by a Google fellowship in Machine Learning.
Web site: https://lucas.maystre.ch.