学会适应：通过认知感知探索实现自我改进的网页智能体
Learning to Adapt: Self-Improving Web Agent via Cognitive-Aware Exploration
对视觉研究者而言，可复用之处在于把 Selector/Predictor/Judger 三角色对抗式自探索与 SCALE-Hop 图探索结合，用于在无专家轨迹条件下自动挖掘 MLLM 智能体的能力边界；这套自改进回路可迁移到需要视觉观测的 GUI/具身导航智能体训练中。
原文依据 we propose SCALE-Hop, a graph exploration strategy that facilitates global planning and helps agents avoid local exploration traps
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论多模态大语言模型（MLLMs）的最新进展推动了网页智能体取得令人瞩目的进步。然而，现有网页智能体往往依赖手工设计的执行流程或昂贵的专家轨迹，限制了其对复杂动态环境的适应能力。为应对这些挑战，我们提出 SCALE（自我认知感知的学习与探索），它利用选择器、预测器与评判器三个对抗性角色，自主发现智能体的能力局限，并通过环境探索拓展其认知边界。此外，我们提出 SCALE-Hop，一种图探索策略，可促进全局规划并帮助智能体避开局部探索陷阱。为进一步支撑学习，我们构建了 SCALE-20k，这是一个从 19 个真实网站采集的大规模数据集，包含多样的任务类型以及由 SCALE 探索轨迹生成的结构化示范。实验结果表明，我们的方法显著提升了多种 MLLM 在各类网页环境中的性能与泛化能力。我们的框架为构建真正自主且自适应的网页智能体提供了一种可扩展、可泛化的方案。
英文原文摘要 1186 字符
Recent advances in Multimodal Large Language Models (MLLMs) have led to promising progress in web agents. However, existing web agents often rely on handcrafted execution pipelines or expensive expert trajectories, limiting their adaptability to complex, dynamic environments. To address these challenges, we propose SCALE (Self-Cognitive-Aware Learning and Exploration), which leverages three adversarial roles--Selector, Predictor, and Judger to autonomously discover the agent's limitations and expand its cognitive boundaries through environmental exploration. Moreover, we propose SCALE-Hop, a graph exploration strategy that facilitates global planning and helps agents avoid local exploration traps. To further support learning, we construct SCALE-20k, a large-scale dataset collected from 19 real-world websites, containing diverse task types and structured demonstrations generated from SCALE's exploration traces. Experimental results show that our approach significantly improves the performance and generalization of multiple MLLMs in various web environments. Our framework offers a scalable and generalizable solution for building truly autonomous and adaptive web agents.