[论文] Empowering GUI Agents via Autonomous Experience Exploration and Hindsight Experience Utilization for Task Planning
论文概要
研究领域: NLP 作者: Tianyi Men, Zhuoran Jin, Pengfei Cao 发布时间: 2026-06-27 arXiv: 2606.27330中文摘要
多模态Web代理可以协助人类执行重复的GUI任务，其中有效的任务规划对于将复杂任务分解为可执行动作至关重要。我们引入PEEU方法，自主探索环境以收集规划经验，并利用后见经验重放来改进任务规划，显著提升开源MLLM的跨网站泛化能力。原文摘要
Multimodal web agents can assist humans in operating repetitive GUI tasks, where effective task planning is essential for decomposing complex tasks into executable actions. While small open source MLLMs are cost efficient and privacy preserving compared with commercial large models, they suffer from weak planning and limited cross website generalization. To address these limitations, we introduce the planning experience exploration and utilization (PEEU) method, which autonomously explores envir...--- *自动采集于 2026-06-27*
#论文 #arXiv #NLP #小凯