REDQT: a method for automated mobile application GUI testing based on deep reinforcement learning algorithms
Publication Date
1-1-2024
Document Type
Article
Publication Title
Service Oriented Computing and Applications
DOI
10.1007/s11761-024-00413-y
Abstract
As mobile applications become increasingly prevalent in daily life, the demand for their functionality and reliability continues to grow. Traditional mobile application testing methods, particularly graphical user interface (GUI) testing, face challenges of limited automation and adaptability.Despite the application of various machine learning approaches to GUI testing, enhancing the utilization of limited component samples in complex mobile application environments remains an overlooked issue in many automated testing methods. This study introduces a mobile application testing method based on deep reinforcement learning, aimed at improving performance and adaptability during the testing process. By integrating the feature recognition capabilities of deep learning with the decision-making mechanisms of reinforcement learning, our method can effectively simulate user operations and identify potential application pitfalls. Initially, the study analyzes the limitations of traditional mobile application testing methods and explores the advantages of deep reinforcement learning in handling complex tasks. Subsequently, we present RedqT: an automated mobile application GUI testing method based on a deep reinforcement learning algorithm (REDQ), aimed at enhancing the utilization of application information through the characteristics of the REDQ algorithm. A study testing 18 open-source Android applications on GitHub demonstrated that our method shows promising performance in terms of code coverage and testing speed.
Keywords
Deep reinforcement learning, Machine learning, Mobile application testing
Department
Computer Engineering
Recommended Citation
Fengyu Wang, Chuanqi Tao, and Jerry Gao. "REDQT: a method for automated mobile application GUI testing based on deep reinforcement learning algorithms" Service Oriented Computing and Applications (2024). https://doi.org/10.1007/s11761-024-00413-y