Multi-intention Inverse Q-learning for Interpretable Behavior Representation

Fuente: arXiv
Guardado en:
Detalles Bibliográficos
Autores principales: Zhu, Hao, De La Crompe, Brice, Kalweit, Gabriel, Schneider, Artur, Kalweit, Maria, Diester, Ilka, Boedecker, Joschka
Formato: Preprint
Publicado: 2023
Materias:
Acceso en línea:
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866914945229127680
author Zhu, Hao
De La Crompe, Brice
Kalweit, Gabriel
Schneider, Artur
Kalweit, Maria
Diester, Ilka
Boedecker, Joschka
author_facet Zhu, Hao
De La Crompe, Brice
Kalweit, Gabriel
Schneider, Artur
Kalweit, Maria
Diester, Ilka
Boedecker, Joschka
contents In advancing the understanding of natural decision-making processes, inverse reinforcement learning (IRL) methods have proven instrumental in reconstructing animal's intentions underlying complex behaviors. Given the recent development of a continuous-time multi-intention IRL framework, there has been persistent inquiry into inferring discrete time-varying rewards with IRL. To address this challenge, we introduce the class of hierarchical inverse Q-learning (HIQL) algorithms. Through an unsupervised learning process, HIQL divides expert trajectories into multiple intention segments, and solves the IRL problem independently for each. Applying HIQL to simulated experiments and several real animal behavior datasets, our approach outperforms current benchmarks in behavior prediction and produces interpretable reward functions. Our results suggest that the intention transition dynamics underlying complex decision-making behavior is better modeled by a step function instead of a smoothly varying function. This advancement holds promise for neuroscience and cognitive science, contributing to a deeper understanding of decision-making and uncovering underlying brain mechanisms.
format Preprint
id arxiv_https___arxiv_org_abs_2311_13870
institution arXiv
publishDate 2023
record_format arxiv
spellingShingle Multi-intention Inverse Q-learning for Interpretable Behavior Representation
Zhu, Hao
De La Crompe, Brice
Kalweit, Gabriel
Schneider, Artur
Kalweit, Maria
Diester, Ilka
Boedecker, Joschka
Machine Learning
Neurons and Cognition
In advancing the understanding of natural decision-making processes, inverse reinforcement learning (IRL) methods have proven instrumental in reconstructing animal's intentions underlying complex behaviors. Given the recent development of a continuous-time multi-intention IRL framework, there has been persistent inquiry into inferring discrete time-varying rewards with IRL. To address this challenge, we introduce the class of hierarchical inverse Q-learning (HIQL) algorithms. Through an unsupervised learning process, HIQL divides expert trajectories into multiple intention segments, and solves the IRL problem independently for each. Applying HIQL to simulated experiments and several real animal behavior datasets, our approach outperforms current benchmarks in behavior prediction and produces interpretable reward functions. Our results suggest that the intention transition dynamics underlying complex decision-making behavior is better modeled by a step function instead of a smoothly varying function. This advancement holds promise for neuroscience and cognitive science, contributing to a deeper understanding of decision-making and uncovering underlying brain mechanisms.
title Multi-intention Inverse Q-learning for Interpretable Behavior Representation
topic Machine Learning
Neurons and Cognition
url https://arxiv.org/abs/2311.13870