[1901.10995] Go-Explore: a New Approach for Hard-Exploration Problems

A grand challenge in reinforcement learning is intelligent exploration, especially when rewards are sparse or deceptive. Two Atari games serve as benchmarks for such hard-exploration domains: Montezuma’s Revenge and Pi…

\nobibliography * Go-Explore: a New Approach for Hard-Exploration Problems Adrien Ecoffet Joost Huizinga Joel Lehman Kenneth O. Stanley* Jeff Clune* Uber AI Labs San Francisco, CA 94103 adrienecoffet,joost.hui,jclune@gmail.com *Co-senior authors Abstract A grand challenge in reinforcement learning is intelligent exploration, especially when rewards are sparse or deceptive. Two Atari games serve as benchmarks for such hard-exploration domains: Montezuma’s Revenge and Pitfall. On both games, current RL algorithms perform poorly, even those with intrinsic motivation, which is the dominant method

related reading

• go-explore-nature.pdfadrien.ecoffet.com
• A (Long) Peek into Reinforcement Learning | Lil'Loglilianweng.github.io
• Exploration Strategies in Deep Reinforcement Learning | Lil'Loglilianweng.github.io
• Debugging Reinforcement Learning Systemsandyljones.com
• Exploration for the Efficient Deployment of Reinforcement Learning Agentsopenreview.net
• Reinforcement Learning via Implicit Imitation Guidancearxiv.org
• How to Explore to Scale RL Training of LLMs on Hard Problems? – Machine Learning Blog | ML@CMU | Carnegie Mellon Universityblog.ml.cmu.edu
• Deep Reinforcement Learning: Pong from Pixelskarpathy.github.io
• Learning Beyond Gradientstrinkle23897.github.io
• Deep Reinforcement Learning Doesn't Work Yetalexirpan.com
• Key Papers in Deep RL - Spinning Up documentationspinningup.openai.com
• Multi-task curriculum learning in a complex, visual,hard-exploration domain: Minecraftarxiv.org