Saved in:
Bibliographic Details
Main Authors: Cho, Geonwoo, Im, Jaegyun, Lee, Jihwan, Yi, Hojun, Kim, Sejin, Kim, Sundong
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2506.19997
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866911516637265920
author Cho, Geonwoo
Im, Jaegyun
Lee, Jihwan
Yi, Hojun
Kim, Sejin
Kim, Sundong
author_facet Cho, Geonwoo
Im, Jaegyun
Lee, Jihwan
Yi, Hojun
Kim, Sejin
Kim, Sundong
contents Generalizing deep reinforcement learning agents to unseen environments remains a significant challenge. One promising solution is Unsupervised Environment Design (UED), a co-evolutionary framework in which a teacher adaptively generates tasks with high learning potential, while a student learns a robust policy from this evolving curriculum. Existing UED methods typically measure learning potential via regret, the gap between optimal and current performance, approximated solely by value-function loss. Building on these approaches, we introduce the transition-prediction error as an additional term in our regret approximation. To capture how training on one task affects performance on others, we further propose a lightweight metric called Co-Learnability. By combining these two measures, we present Transition-aware Regret Approximation with Co-learnability for Environment Design (TRACED). Empirical evaluations show that TRACED produces curricula that improve zero-shot generalization over strong baselines across multiple benchmarks. Ablation studies confirm that the transition-prediction error drives rapid complexity ramp-up and that Co-Learnability delivers additional gains when paired with the transition-prediction error. These results demonstrate how refined regret approximation and explicit modeling of task relationships can be leveraged for sample-efficient curriculum design in UED. Project Page: https://geonwoo.me/traced/
format Preprint
id arxiv_https___arxiv_org_abs_2506_19997
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle TRACED: Transition-aware Regret Approximation with Co-learnability for Environment Design
Cho, Geonwoo
Im, Jaegyun
Lee, Jihwan
Yi, Hojun
Kim, Sejin
Kim, Sundong
Machine Learning
Artificial Intelligence
Generalizing deep reinforcement learning agents to unseen environments remains a significant challenge. One promising solution is Unsupervised Environment Design (UED), a co-evolutionary framework in which a teacher adaptively generates tasks with high learning potential, while a student learns a robust policy from this evolving curriculum. Existing UED methods typically measure learning potential via regret, the gap between optimal and current performance, approximated solely by value-function loss. Building on these approaches, we introduce the transition-prediction error as an additional term in our regret approximation. To capture how training on one task affects performance on others, we further propose a lightweight metric called Co-Learnability. By combining these two measures, we present Transition-aware Regret Approximation with Co-learnability for Environment Design (TRACED). Empirical evaluations show that TRACED produces curricula that improve zero-shot generalization over strong baselines across multiple benchmarks. Ablation studies confirm that the transition-prediction error drives rapid complexity ramp-up and that Co-Learnability delivers additional gains when paired with the transition-prediction error. These results demonstrate how refined regret approximation and explicit modeling of task relationships can be leveraged for sample-efficient curriculum design in UED. Project Page: https://geonwoo.me/traced/
title TRACED: Transition-aware Regret Approximation with Co-learnability for Environment Design
topic Machine Learning
Artificial Intelligence
url https://arxiv.org/abs/2506.19997