Optimistic Simulated Exploration as an Incentive for Real Exploration

TL;DR

This paper introduces a method that uses optimistic simulated exploration to identify promising paths, reducing the need for extensive real-world exploration in environments with many states.

Contribution

It proposes a novel approach combining optimistic simulated exploration with real exploration to improve efficiency in environments with unlimited states.

Findings

Reduces real exploration needs significantly

Effective in environments with large or infinite state spaces

Improves exploration efficiency over traditional methods

Abstract

Many reinforcement learning exploration techniques are overly optimistic and try to explore every state. Such exploration is impossible in environments with the unlimited number of states. I propose to use simulated exploration with an optimistic model to discover promising paths for real exploration. This reduces the needs for the real exploration.