# Hill Climbing on Value Estimates for Search-control in Dyna

## TL;DR

This paper introduces HC-Dyna, a novel search-control method for model-based RL that uses hill climbing on value estimates to improve sample efficiency, demonstrating significant gains in classical domains.

## Contribution

It proposes a new search-control mechanism using hill climbing on value functions, with a derived natural gradient algorithm and empirical validation in RL tasks.

## Key findings

- HC-Dyna improves sample efficiency in classical RL domains.
- Using hill climbing on value estimates from low to high regions benefits search-control.
- The approach connects to Langevin dynamics, offering a theoretical foundation.

## Abstract

Dyna is an architecture for model-based reinforcement learning (RL), where simulated experience from a model is used to update policies or value functions. A key component of Dyna is search-control, the mechanism to generate the state and action from which the agent queries the model, which remains largely unexplored. In this work, we propose to generate such states by using the trajectory obtained from Hill Climbing (HC) the current estimate of the value function. This has the effect of propagating value from high-value regions and of preemptively updating value estimates of the regions that the agent is likely to visit next. We derive a noisy projected natural gradient algorithm for hill climbing, and highlight a connection to Langevin dynamics. We provide an empirical demonstration on four classical domains that our algorithm, HC-Dyna, can obtain significant sample efficiency improvements. We study the properties of different sampling distributions for search-control, and find that there appears to be a benefit specifically from using the samples generated by climbing on current value estimates from low-value to high-value region.