Neural Information Processing Systems http://nips.cc/

Neural Information Processing Systems http://nips.cc/

However, in many real-world applications, collecting sufficient exploratory interactions is usually impractical, because online datacollection canbecostlyorevendangerous, suchasinhealthcare [4]andautonomous driving [5]. To address this challenge, offline RL [6, 7] develops a new learning paradigm that trains RL agents only with pre-collected offline datasets and thus can abstract away from the cost of online exploration [8-17].

However, in many real-world applications, collecting sufficient exploratory interactions is usually impractical, because online datacollection canbecostlyorevendangerous, suchasinhealthcare [4]andautonomous driving [5]. To address this challenge, offline RL [6, 7] develops a new learning paradigm that trains RL agents only with pre-collected offline datasets and thus can abstract away from the cost of online exploration [8-17].

Research in artificial intelligence, and particularly deep reinforcement learning (RL), relies on evaluating aggregate performance on a diverse suite of tasks to assess progress.

This paper investigates posterior sampling algorithms for competitive reinforcement learning (RL) in the context of general function approximations.

In human pedagogy, teachers and students can interact adaptively to maximize communication efficiency. Theteacher adjusts herteaching method fordifferent students, and the student, after getting familiar with the teacher's instruction mechanism,caninfertheteacher'sintentiontolearnfaster.

Our algorithm SPOT consists of two stages, namely VAE training and policytraining.