Goto

Collaborating Authors

 Learning Graphical Models

Reinforcement learning from Human Feedback (RLHF) learns from preference signals, while standard Reinforcement Learning (RL) directly learns from reward

Neural Information Processing Systems

The latter case can be further reduced to adversarial MDP when preferences only depend on the final state.

efaf1c9726648c8ba363a5c927440529-Paper-Conference.pdf

bd8b52c2fefdb37e3b3953a37408e9dc-Paper-Conference.pdf

Learning Causal Models under Independent Changes

Neural Information Processing Systems

In many scientific applications, we observe a system in different conditions in which its components may change, rather than in isolation.

bcd11db0b26d8fc2266b91d3ff982ed1-Paper-Conference.pdf

bc82dbfbfa43232be85b8d9838f49c3e-Paper-Conference.pdf

DIFFER: Decomposing Individual Reward for Fair Experience Replay in Multi-Agent Reinforcement Learning

Neural Information Processing Systems

The code is available in https://github.com/cathyhxh/DIFFER.

OPERA: Automatic Offline Policy Evaluation with Re-weighted Aggregates of Multiple Estimators

Neural Information Processing Systems

Unfortunately, choosing the best OPE algorithm for each task and domain is still unclear.

eb5254c4ee813d05af9c098f2d9c5708-Paper-Conference.pdf

Integrating Suboptimal Human Knowledge with Hierarchical Reinforcement Learning for Large-Scale Multiagent Systems

Neural Information Processing Systems

While agents' learning is data-driven, sampling from millions