Learning Top-K Subtask Planning Tree Based on Discriminative Representation Pretraining for Decision-making

doi:10.1007/s11633-023-1483-z

CASIA OpenIR > 学术期刊 > Machine Intelligence Research

	Learning Top-K Subtask Planning Tree Based on Discriminative Representation Pretraining for Decision-making
	Jingqing Ruan 1,2; Kaishen Wang 1,3; Qingyang Zhang 1,2; Dengpeng Xing 1,3; Bo Xu 1,3
发表期刊	Machine Intelligence Research
ISSN	2731-538X
	2024
卷号	21 期号:4 页码:782-800
摘要	Decomposing complex real-world tasks into simpler subtasks and devising a subtask execution plan is critical for humans to achieve effective decision-making. However, replicating this process remains challenging for AI agents and naturally raises two questions: 1) How to extract discriminative knowledge representation from priors? 2) How to develop a rational plan to decompose complex problems? To address these issues, we introduce a groundbreaking framework that incorporates two main contributions. First, our multiple-encoder and individual-predictor regime goes beyond traditional architectures to extract nuanced task-specific dynamics from datasets, enriching the feature space for subtasks. Second, we innovate in planning by introducing a top- subtask planning tree generated through an attention mechanism, which allows for dynamic adaptability and forward-looking decision-making. Our framework is empirically validated against challenging benchmarks BabyAI including multiple combinatorially rich synthetic tasks (e.g., GoToSeq, SynthSeq, BossLevel), where it not only outperforms competitive baselines but also demonstrates superior adaptability and effectiveness in complex task decomposition.
关键词	Reinforcement learning representation learning subtask planning task decomposition pretraining.
DOI	10.1007/s11633-023-1483-z
引用统计
文献类型	期刊论文
条目标识符	http://ir.ia.ac.cn/handle/173211/58572
专题	学术期刊_Machine Intelligence Research
作者单位	1.Institute of Automation, Chinese Academy of Sciences, Beijing 100190, China 2.School of Future Technology, University of Chinese Academy of Sciences, Beijing 100049, China 3.School of Artificial Intelligence, University of Chinese Academy of Sciences, Beijing 100049, China
第一作者单位	中国科学院自动化研究所
推荐引用方式 GB/T 7714	Jingqing Ruan, Kaishen Wang, Qingyang Zhang,et al. Learning Top-K Subtask Planning Tree Based on Discriminative Representation Pretraining for Decision-making[J]. Machine Intelligence Research,2024,21(4):782-800.
APA	Jingqing Ruan, Kaishen Wang, Qingyang Zhang, Dengpeng Xing,& Bo Xu.(2024).Learning Top-K Subtask Planning Tree Based on Discriminative Representation Pretraining for Decision-making.Machine Intelligence Research,21(4),782-800.
MLA	Jingqing Ruan,et al."Learning Top-K Subtask Planning Tree Based on Discriminative Representation Pretraining for Decision-making".Machine Intelligence Research 21.4(2024):782-800.

条目包含的文件
文件名称/大小	文献类型	版本类型	开放类型	使用许可
MIR-2023-04-059.pdf（4577KB）	期刊论文	出版稿	开放获取	CC BY-NC-SA	浏览下载