Reinforcement Learning
A Structured Prediction Approach for Generalization in Cooperative Multi-Agent Reinforcement Learning
Nicolas Carion, Nicolas Usunier, Gabriel Synnaeve, Alessandro Lazaric
Neural Information Processing Systems http://nips.cc/
Reviewer 1: Q1: I wonder if their analysis tricks of AC/NAC when applied to PG methods improve their guarantees
If their analysis tricks do improve PG guarantees, how does it compare then? Reviewer 2: Q1: It would be interesting to complement the theoretical results with empirical results in toy problem. We are working on experiments and will add these results to the revision. Q2: For the error term that disappears with a larger mini-batch (line 211). A2: Y es, this error term should be called as variance error.
Memory Based Trajectory-conditioned Policies for Learning from Sparse Rewards
Recent work demonstrated that using a memory buffer of previous successful trajectories can result in more effective policies. However, existing methods may overly exploit past successful experiences, which can encourage the agent to adopt sub-optimal and myopic behaviors.
Mapping State Space using Landmarks for Universal Goal Reaching
Zhiao Huang, Fangchen Liu, Hao Su
Neural Information Processing Systems http://nips.cc/
Munchausen Reinforcement Learning
Bootstrapping is a core mechanism in Reinforcement Learning (RL).
