Time: Wednesday 24-Jun-2020 16:00 (This is a past event.)
Motivation / Abstract
Collaborative filtering is widely used in modern recommender systems. In this paper we investigate new methods for training collaborative filtering models based on actor-critic reinforcement learning, to directly optimize the non-differentiable quality metrics of interest. Empirically, we show that the proposed methods outperform several state-of-the-art baselines, including recently-proposed deep learning approaches, on three large-scale real-world datasets.
1) VAE (actor) accompanied with the critic can accurately predict latent representation for users and items 2) It is not typical RL nor typical RecSys therefore the evaluation part had to be customized to fit the problem 3) Using the proposed method, the rating/utility matrix sparsity problem is overcame