AI Testing 2019 will feature 4 special tracks, in addition to research track, organized by technical leaders of their community, with the the following characteristics:
4 Special Tracks
Organizer: Zhi Quan (George) Zhou, University of Wollongong, Australia
Metamorphic testing (MT) is a software quality assurance paradigm that can effectively alleviate the test oracle problem and enable automated test case generation. In recent years, MT has been increasingly used for the testing of artificial intelligence (AI) systems, including various machine learning applications and autonomous systems. Furthermore, machine learning techniques have also been applied to enhance MT. This special track will bring together researchers and practitioners in academia and industry to discuss research results and experiences that explore the interplay between MT and AI. We invite original submissions on, but not limited to, the following topics: MT of AI systems, AI techniques for MT, and the use of metamorphic relations for enhancing machine learning.
Organizer: Helge Spieker, Simula Research Laboratory, Norway
Improving the Testing Process through Machine Learning. The goal of this special track is to highlight advances in the application of machine learning techniques to increase the efficiency of software testing. In modern software engineering, testing often consists of repeatedly executing the system-under-test, either in continuous integration, but also in extended stand-alone testing of a software system. We are looking for solutions that introduce machine learning techniques into traditional and established software testing techniques, for example stress/mutation/metamorphic testing.
Organizer: Nadjib Lazaar, LIRMM, France
Recent years have seen a revolutionary improvement in the efficiency and expressive power of constraint solvers, such as Boolean SAT solvers, Satisfiability Modulo Theory (SMT) and Constraint Programming (CP) solvers, with a consequent impact on Software Testing, Verification and Analysis techniques. Constraint-based testing and symbolic-execution techniques have emerged as strong methods in automatic test case generation, oracle checking, static analysis, bounded and abstract model-checking, fault detection, localization and correction security testing, etc. A great deal of work remains to determine on how to best tune and extend these solvers to meet the needs of particular applications.
Organizer: Sébastien Bardin, CEA, France
Cybersecurity is a major concern of today's societies. The 2016 DARPA Cyber Grand Challenge highlighted that automated AI-based cyber-reasoning systems can be effective in particular areas of security testing, while AI-based systems have recently been proven vulnerable to attacks based on automated generation of adversarial input. The "AI and security testing" track will gather participants interested on these interplays between AI techniques and security testing.