Engineering Complex Software: Mutation analysis, Test Generation, and Neural Networks Repair
Thomas Laurent, NII, Japan
3pm 4th Jul 2023
Abstract
As software systems become ubiquitous, and the impact of software faults becomes larger, so does the necessity to ensure the quality of tests. Software Engineering, and in particular Software Testing help us ensure the conformity of software systems to given requirements, and limit the risk of faults. In this talk, I will discuss work around 1) test assessment, evaluating the quality of tests through mutation analysis; 2) test generation, to obtain additional tests; and 3) repair of Neural Networks based systems when a fault is detected. I will present these works through the lens of the ever growing scale and complexity of modern software systems.
Short Bio
Thomas Laurent is a Japanese Society for the Promotion of Science (JSPS) research fellow hosted by the National Institute of Informatics, Tokyo, Japan. He is part of the Engineerable AI (eAI, https://engineerable.ai/en/) project, and of Trinity College Dublin’s Complex Software Laboratory (https://csl.scss.tcd.ie/) , where he explores problems of Software Engineering and Testing for modern complex and smart systems. He received his PhD in Computer Science from University College Dublin in 2021, for which he focused on applying and extending mutation analysis to various domains.