LearnAct: Few-Shot Mobile GUI Agent with a Unified Demonstration Benchmark

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Liu, Guangyi, Zhao, Pengxiang, Liu, Liang, Chen, Zhiming, Chai, Yuxiang, Ren, Shuai, Wang, Hao, He, Shibo, Meng, Wenchao
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866909584306733056
author Liu, Guangyi
Zhao, Pengxiang
Liu, Liang
Chen, Zhiming
Chai, Yuxiang
Ren, Shuai
Wang, Hao
He, Shibo
Meng, Wenchao
author_facet Liu, Guangyi
Zhao, Pengxiang
Liu, Liang
Chen, Zhiming
Chai, Yuxiang
Ren, Shuai
Wang, Hao
He, Shibo
Meng, Wenchao
contents Mobile GUI agents show promise in automating tasks but face generalization challenges in diverse real-world scenarios. Traditional approaches using pre-training or fine-tuning with massive datasets struggle with the diversity of mobile applications and user-specific tasks. We propose enhancing mobile GUI agent capabilities through human demonstrations, focusing on improving performance in unseen scenarios rather than pursuing universal generalization through larger datasets. To realize this paradigm, we introduce LearnGUI, the first comprehensive dataset specifically designed for studying demonstration-based learning in mobile GUI agents, comprising 2,252 offline tasks and 101 online tasks with high-quality human demonstrations. We further develop LearnAct, a sophisticated multi-agent framework that automatically extracts knowledge from demonstrations to enhance task completion. This framework integrates three specialized agents: DemoParser for knowledge extraction, KnowSeeker for relevant knowledge retrieval, and ActExecutor for demonstration-enhanced task execution. Our experimental results show significant performance gains in both offline and online evaluations. In offline assessments, a single demonstration improves model performance, increasing Gemini-1.5-Pro's accuracy from 19.3% to 51.7%. In online evaluations, our framework enhances UI-TARS-7B-SFT's task success rate from 18.1% to 32.8%. LearnAct framework and LearnGUI benchmark establish demonstration-based learning as a promising direction for more adaptable, personalized, and deployable mobile GUI agents.
format Preprint
id arxiv_https___arxiv_org_abs_2504_13805
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle LearnAct: Few-Shot Mobile GUI Agent with a Unified Demonstration Benchmark
Liu, Guangyi
Zhao, Pengxiang
Liu, Liang
Chen, Zhiming
Chai, Yuxiang
Ren, Shuai
Wang, Hao
He, Shibo
Meng, Wenchao
Human-Computer Interaction
Mobile GUI agents show promise in automating tasks but face generalization challenges in diverse real-world scenarios. Traditional approaches using pre-training or fine-tuning with massive datasets struggle with the diversity of mobile applications and user-specific tasks. We propose enhancing mobile GUI agent capabilities through human demonstrations, focusing on improving performance in unseen scenarios rather than pursuing universal generalization through larger datasets. To realize this paradigm, we introduce LearnGUI, the first comprehensive dataset specifically designed for studying demonstration-based learning in mobile GUI agents, comprising 2,252 offline tasks and 101 online tasks with high-quality human demonstrations. We further develop LearnAct, a sophisticated multi-agent framework that automatically extracts knowledge from demonstrations to enhance task completion. This framework integrates three specialized agents: DemoParser for knowledge extraction, KnowSeeker for relevant knowledge retrieval, and ActExecutor for demonstration-enhanced task execution. Our experimental results show significant performance gains in both offline and online evaluations. In offline assessments, a single demonstration improves model performance, increasing Gemini-1.5-Pro's accuracy from 19.3% to 51.7%. In online evaluations, our framework enhances UI-TARS-7B-SFT's task success rate from 18.1% to 32.8%. LearnAct framework and LearnGUI benchmark establish demonstration-based learning as a promising direction for more adaptable, personalized, and deployable mobile GUI agents.
title LearnAct: Few-Shot Mobile GUI Agent with a Unified Demonstration Benchmark
topic Human-Computer Interaction
url https://arxiv.org/abs/2504.13805