About this document

Self-Play RL for Protein Engineering by 1813975862yvyv is a document available to read on EtoBox.

The document describes a new machine learning framework called EvoPlay that uses self-play reinforcement learning to guide protein engineering. EvoPlay treats protein sequence optimization as a game where mutations are actions, and uses a policy-value neural network and Monte Carlo tree search to iteratively sample sequences and train the network. This allows EvoPlay to efficiently search the vast protein design space. The authors evaluate EvoPlay on several directed evolution tasks, where it discovers vari

Author

1813975862yvyv

Language

EN