Composing Task Knowledge with Modular Successor Feature Approximators

Recently, the Successor Features and Generalized Policy Improvement (SF&GPI) framework has been proposed as a method for learning, composing, and transferring predictive knowledge and behavior. SF&GPI works by having an agent learn predictive representations (SFs) that can be combined for tr...

Full description

Saved in:
Bibliographic Details
Main Authors Carvalho, Wilka, Filos, Angelos, Lewis, Richard L, lee, Honglak, Singh, Satinder
Format Journal Article
LanguageEnglish
Published 28.01.2023
Subjects
Computer Science - Artificial Intelligence
Computer Science - Learning
Online AccessGet full text

Cover

More Information
Summary:Recently, the Successor Features and Generalized Policy Improvement (SF&GPI) framework has been proposed as a method for learning, composing, and transferring predictive knowledge and behavior. SF&GPI works by having an agent learn predictive representations (SFs) that can be combined for transfer to new tasks with GPI. However, to be effective this approach requires state features that are useful to predict, and these state-features are typically hand-designed. In this work, we present a novel neural network architecture, "Modular Successor Feature Approximators" (MSFA), where modules both discover what is useful to predict, and learn their own predictive representations. We show that MSFA is able to better generalize compared to baseline architectures for learning SFs and modular architectures
DOI:10.48550/arxiv.2301.12305