Hybrid Reinforcement Learning with Expert State Sequences

March 11, 2019 ยท Entered Twilight ยท ๐Ÿ› AAAI Conference on Artificial Intelligence

๐ŸŒ… TWILIGHT: Old Age
Predates the code-sharing era โ€” a pioneer of its time

"Last commit was 6.0 years ago (โ‰ฅ5 year threshold)"

Evidence collected by the PWNC Scanner

Repo contents: .DS_Store, .gitattributes, .gitignore, LICENSE, README.md, algo, arguments.py, distributions.py, dual_model.py, enjoy.py, envs.py, evaluate.py, main.py, model.py, policies.py, storage.py, utils.py, visualize.py

Authors Xiaoxiao Guo, Shiyu Chang, Mo Yu, Gerald Tesauro, Murray Campbell arXiv ID 1903.04110 Category cs.LG: Machine Learning Cross-listed cs.AI, stat.ML Citations 35 Venue AAAI Conference on Artificial Intelligence Repository https://github.com/XiaoxiaoGuo/tensor4rl โญ 3 Last Checked 1 month ago
Abstract
Existing imitation learning approaches often require that the complete demonstration data, including sequences of actions and states, are available. In this paper, we consider a more realistic and difficult scenario where a reinforcement learning agent only has access to the state sequences of an expert, while the expert actions are unobserved. We propose a novel tensor-based model to infer the unobserved actions of the expert state sequences. The policy of the agent is then optimized via a hybrid objective combining reinforcement learning and imitation learning. We evaluated our hybrid approach on an illustrative domain and Atari games. The empirical results show that (1) the agents are able to leverage state expert sequences to learn faster than pure reinforcement learning baselines, (2) our tensor-based action inference model is advantageous compared to standard deep neural networks in inferring expert actions, and (3) the hybrid policy optimization objective is robust against noise in expert state sequences.
Community shame:
Not yet rated
Community Contributions

Found the code? Know the venue? Think something is wrong? Let us know!

๐Ÿ“œ Similar Papers

In the same crypt โ€” Machine Learning