Automated Testing and Analysis for Dependable AI-enabled Software and Systems
摘要截稿:
全文截稿: 2024-04-07
影响因子: 2.45
期刊难度:
CCF分类: B类
中科院JCR分区:
• 大类 : 计算机科学 - 2区
• 小类 : 计算机:软件工程 - 2区
• 小类 : 计算机:理论方法 - 2区
Overview
The advancements in Artificial Intelligence (AI) and its integration into various domains have led to the development of AI-enabled software and systems that offer unprecedented capabilities. Technologies ranging from computer vision to natural language processing, from speech recognition to recommender systems enhance modern software and systems with the aim of providing innovative services, as well as rich and customized experiences to the users. Such technologies are also changing the software and system engineering and development methods and tools, especially quality assurance methods that require deep restructuring due to the inherent differences between AI and traditional software.
------------
AI-enabled software and systems are often large-scale driven by data, and more complex than traditional software and systems. They are typically heterogeneous, autonomous, and probabilistic in nature. They also lack of transparent understanding of their internal mechanics. Furthermore, they are typically optimized and trained for specific tasks and, as such, may fail to generalize their knowledge to other situations that often emerge in dynamic environments. These systems strongly demand safety, trustworthiness, security, and other dependability aspects. High-quality data and AI components shall be safely integrated, verified, maintained, and evolved. In fact, the potential impact of a failure, or a service interruption, cannot be tolerated in business-critical applications (e.g., chatbots and virtual assistants, facial recognition for authentication and security, industrial robots) or safety-critical applications (e.g., autonomous drones, collaborative robots, self-driving cars and autonomous vehicles for transportation).
------------
The scientific community is hence studying new cost-effective verification and validation techniques tailored to these systems. In particular, automated testing and analysis is a very active area that has led to notable advances to realize the promise of dependable AI-enabled software and systems.
------------
This special issue welcomes contributions regarding approaches, techniques, tools, and experience reports about adopting, creating, and improving automated testing and analysis of AI-enabled software and systems with a special focus on dependability aspects, such as reliability, safety, security, resilience, scalability, usability, trustworthiness, and compliance to standards.
--