Advantage Amplification in Slowly Evolving Latent-State Environments
Advantage Amplification in Slowly Evolving Latent-State Environments
Martin Mladenov, Ofer Meshi, Jayden Ooi, Dale Schuurmans, Craig Boutilier
Proceedings of the Twenty-Eighth International Joint Conference on Artificial Intelligence
Main track. Pages 3165-3172.
https://doi.org/10.24963/ijcai.2019/439
Latent-state environments with long horizons, such as those faced by recommender systems, pose significant challenges for reinforcement learning (RL). In this work, we identify and analyze several key hurdles for RL in such environments, including belief state error and small action advantage. We develop a general principle called advantage amplification that an overcome these hurdles through the use of temporal abstraction. We propose several aggregation methods and prove they induce amplification in certain settings. We also bound the loss in optimality incurred by our methods in environments where latent state evolves slowly and demonstrate their performance empirically in a stylized user-modeling task.
Keywords:
Machine Learning: Reinforcement Learning
Machine Learning: Time-series;Data Streams
Humans and AI: Personalization and User Modeling
Machine Learning: Recommender Systems