Although several studies are carried out on Machine Learning based systems, their reliability engineering remains to be thoroughly explored.
Machine Learning based systems have become increasingly popular in various areas that involve advanced data analysis. Concerning the usage of Machine Learning based systems in critical domains, where a small error may lead to a disaster such as autonomous vehicles and smart healthcare, there is a significant need for reliable Machine Learning based systems which can generate trustworthy results. What makes the reliability engineering of Machine Learning based systems a challenge is its requirement of expertise in both Software Engineering and Machine Learning at the same time. Besides, the reliability engineering techniques working well for traditional software systems are insufficient and ineffective for Machine Learning based systems. We aim to characterize bugs and quality attributes in these systems.
Team
References