VDSC: Enhancing Exploration Timing with Value Discrepancy and State Counts

Fuente: arXiv
Guardado en:
Detalles Bibliográficos
Autores principales: Captari, Marius, Sasso, Remo, Sabatelli, Matthia
Formato: Preprint
Publicado: 2024
Materias:
Acceso en línea:
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866910384102834176
author Captari, Marius
Sasso, Remo
Sabatelli, Matthia
author_facet Captari, Marius
Sasso, Remo
Sabatelli, Matthia
contents Despite the considerable attention given to the questions of \textit{how much} and \textit{how to} explore in deep reinforcement learning, the investigation into \textit{when} to explore remains relatively less researched. While more sophisticated exploration strategies can excel in specific, often sparse reward environments, existing simpler approaches, such as $ε$-greedy, persist in outperforming them across a broader spectrum of domains. The appeal of these simpler strategies lies in their ease of implementation and generality across a wide range of domains. The downside is that these methods are essentially a blind switching mechanism, which completely disregards the agent's internal state. In this paper, we propose to leverage the agent's internal state to decide \textit{when} to explore, addressing the shortcomings of blind switching mechanisms. We present Value Discrepancy and State Counts through homeostasis (VDSC), a novel approach for efficient exploration timing. Experimental results on the Atari suite demonstrate the superiority of our strategy over traditional methods such as $ε$-greedy and Boltzmann, as well as more sophisticated techniques like Noisy Nets.
format Preprint
id arxiv_https___arxiv_org_abs_2403_17542
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle VDSC: Enhancing Exploration Timing with Value Discrepancy and State Counts
Captari, Marius
Sasso, Remo
Sabatelli, Matthia
Machine Learning
Artificial Intelligence
Despite the considerable attention given to the questions of \textit{how much} and \textit{how to} explore in deep reinforcement learning, the investigation into \textit{when} to explore remains relatively less researched. While more sophisticated exploration strategies can excel in specific, often sparse reward environments, existing simpler approaches, such as $ε$-greedy, persist in outperforming them across a broader spectrum of domains. The appeal of these simpler strategies lies in their ease of implementation and generality across a wide range of domains. The downside is that these methods are essentially a blind switching mechanism, which completely disregards the agent's internal state. In this paper, we propose to leverage the agent's internal state to decide \textit{when} to explore, addressing the shortcomings of blind switching mechanisms. We present Value Discrepancy and State Counts through homeostasis (VDSC), a novel approach for efficient exploration timing. Experimental results on the Atari suite demonstrate the superiority of our strategy over traditional methods such as $ε$-greedy and Boltzmann, as well as more sophisticated techniques like Noisy Nets.
title VDSC: Enhancing Exploration Timing with Value Discrepancy and State Counts
topic Machine Learning
Artificial Intelligence
url https://arxiv.org/abs/2403.17542