A0510
Title: Recommender systems, bandits and Bayesian neural networks
Authors: Simen Eide - University of Oslo / Schibsted (Norway) [presenting]
Abstract: Internet platforms consist of millions or billions of different items that users can consume. To help users navigate this landscape, recommender systems have become an important component in many platforms. The aim of a recommender system is to suggest the most relevant content on the platform to the user based on previous interactions the user has done with the platform. A model used in recommender systems faces multiple sources of uncertainty: There are limited interactions per user, the signals a user makes may be noisy and not always reflect her preferences, and new items may be introduced to the platform giving few signals on these items as well. The focus will be on model uncertainty and decision making in the recommender systems. We will discuss various ways to quantify, reduce and exploit these uncertainties through the use of Bayesian neural networks, hierarchical priors and different recommender strategies.
