Cracking the Code of Action: a Generative Approach to Affordances for Reinforcement Learning

Fuente: arXiv
Guardado en:
Detalles Bibliográficos
Autores principales: Cherif, Lynn, Kondrup, Flemming, Venuto, David, Anand, Ankit, Precup, Doina, Khetarpal, Khimya
Formato: Preprint
Publicado: 2025
Materias:
Acceso en línea:
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866916705606828032
author Cherif, Lynn
Kondrup, Flemming
Venuto, David
Anand, Ankit
Precup, Doina
Khetarpal, Khimya
author_facet Cherif, Lynn
Kondrup, Flemming
Venuto, David
Anand, Ankit
Precup, Doina
Khetarpal, Khimya
contents Agents that can autonomously navigate the web through a graphical user interface (GUI) using a unified action space (e.g., mouse and keyboard actions) can require very large amounts of domain-specific expert demonstrations to achieve good performance. Low sample efficiency is often exacerbated in sparse-reward and large-action-space environments, such as a web GUI, where only a few actions are relevant in any given situation. In this work, we consider the low-data regime, with limited or no access to expert behavior. To enable sample-efficient learning, we explore the effect of constraining the action space through $\textit{intent-based affordances}$ -- i.e., considering in any situation only the subset of actions that achieve a desired outcome. We propose $\textbf{Code as Generative Affordances}$ $(\textbf{$\texttt{CoGA}$})$, a method that leverages pre-trained vision-language models (VLMs) to generate code that determines affordable actions through implicit intent-completion functions and using a fully-automated program generation and verification pipeline. These programs are then used in-the-loop of a reinforcement learning agent to return a set of affordances given a pixel observation. By greatly reducing the number of actions that an agent must consider, we demonstrate on a wide range of tasks in the MiniWob++ benchmark that: $\textbf{1)}$ $\texttt{CoGA}$ is orders of magnitude more sample efficient than its RL agent, $\textbf{2)}$ $\texttt{CoGA}$'s programs can generalize within a family of tasks, and $\textbf{3)}$ $\texttt{CoGA}$ performs better or on par compared with behavior cloning when a small number of expert demonstrations is available.
format Preprint
id arxiv_https___arxiv_org_abs_2504_17282
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Cracking the Code of Action: a Generative Approach to Affordances for Reinforcement Learning
Cherif, Lynn
Kondrup, Flemming
Venuto, David
Anand, Ankit
Precup, Doina
Khetarpal, Khimya
Artificial Intelligence
Agents that can autonomously navigate the web through a graphical user interface (GUI) using a unified action space (e.g., mouse and keyboard actions) can require very large amounts of domain-specific expert demonstrations to achieve good performance. Low sample efficiency is often exacerbated in sparse-reward and large-action-space environments, such as a web GUI, where only a few actions are relevant in any given situation. In this work, we consider the low-data regime, with limited or no access to expert behavior. To enable sample-efficient learning, we explore the effect of constraining the action space through $\textit{intent-based affordances}$ -- i.e., considering in any situation only the subset of actions that achieve a desired outcome. We propose $\textbf{Code as Generative Affordances}$ $(\textbf{$\texttt{CoGA}$})$, a method that leverages pre-trained vision-language models (VLMs) to generate code that determines affordable actions through implicit intent-completion functions and using a fully-automated program generation and verification pipeline. These programs are then used in-the-loop of a reinforcement learning agent to return a set of affordances given a pixel observation. By greatly reducing the number of actions that an agent must consider, we demonstrate on a wide range of tasks in the MiniWob++ benchmark that: $\textbf{1)}$ $\texttt{CoGA}$ is orders of magnitude more sample efficient than its RL agent, $\textbf{2)}$ $\texttt{CoGA}$'s programs can generalize within a family of tasks, and $\textbf{3)}$ $\texttt{CoGA}$ performs better or on par compared with behavior cloning when a small number of expert demonstrations is available.
title Cracking the Code of Action: a Generative Approach to Affordances for Reinforcement Learning
topic Artificial Intelligence
url https://arxiv.org/abs/2504.17282