TestFlow: Advancing Mobile UI Testing through Multi-Step Reinforcement Learning
GUI Agents have demonstrated promising applications in mobile UI testing. However, for complex testing tasks, UI agents tend to fail due to their greedy approach in executing step-by-step operations, leading to error accumulation and neglecting long-horizon dependencies. To address these limitations, we propose TestFlow, a novel multi-modal UI testing model that combines Supervised Fine-Tuning with a Task-aware Reinforcement Learning framework. Our approach implements a two-phase training pipeline designed to optimize long-horizon instruction compliance and complex task completion. Additionally, we develop a tailor-made reward function that integrates both process and outcome rewards to improve the completion rate of multi-step tasks. The experimental results demonstrate that TestFlow significantly outperforms the baseline methods, achieving 33. 69% WTSR and 55. 37% SSR in cross-page test scenarios. These improvements highlight the practical value of TestFlow in addressing the challenges of modern mobile app testing, particularly in industrial settings requiring high adaptability and reliability.
Sat 28 JunDisplayed time zone: Amsterdam, Berlin, Bern, Rome, Stockholm, Vienna change
11:00 - 12:30 | Intelligence and PrivacyEXPRESS at Cosmos 3B Chair(s): Peng Di Ant Group & UNSW Sydney, Puzhuo Liu Ant Group & Tsinghua University | ||
11:00 20mTalk | Patch the Leak: Strengthening CodeLLMs Against Privacy Extraction Threats EXPRESS Yongjian Guo Tsinghua University & Ant Group, Wanlun Ma Swinburne University of Technology, Xi Xiao Tsinghua University, Sheng Wen Swinburne University of Technology, Peng Di Ant Group & UNSW Sydney, Xiaogang Zhu The University of Adelaide | ||
11:20 20mTalk | From Large Language Models to Adversarial Malware: How far are we EXPRESS Shuai He Huazhong University of Science and Technology, Hao Yan Huazhong University of Science and Technology, Wenke Li Huazhong University of Science and Technology, Sheng Hong Huazhong University of Science and Technology, Xiaowei Guo Huazhong University of Science and Technology, Xiaofan Liu Huazhong University of Science and Technology, Cai Fu Huazhong University of Science and Technology | ||
11:40 20mTalk | Towards Source Mapping for Zero-Knowledge Smart Contracts: Design and Preliminary Evaluation EXPRESS Pei Xu University of Technology Sydney, Yulei Sui University of New South Wales, Mark Staples Digital Finance CRC | ||
12:00 20mTalk | TestFlow: Advancing Mobile UI Testing through Multi-Step Reinforcement Learning EXPRESS Xiaoxuan Tang Ant Group, Xinfang Chen Ant Group, Dajun Chen Ant Group, Sheng Zhou Zhejiang University, Wei Jiang Ant Group, Yong Li Ant Group | ||
12:20 10mDay closing | Discussion and Conclusion EXPRESS | ||
Cosmos 3B is the second room in the Cosmos 3 wing.
When facing the main Cosmos Hall, access to the Cosmos 3 wing is on the left, close to the stairs. The area is accessed through a large door with the number “3”, which will stay open during the event.