Lifelong self-adaptation Article Swipe
YOU?
·
· 2022
· Open Access
·
· DOI: https://doi.org/10.1145/3524844.3528052
· OA: W4226183793
In the past years, machine learning (ML) has become a popular approach to\nsupport self-adaptation. While ML techniques enable dealing with several\nproblems in self-adaptation, such as scalable decision-making, they are also\nsubject to inherent challenges. In this paper, we focus on one such challenge\nthat is particularly important for self-adaptation: ML techniques are designed\nto deal with a set of predefined tasks associated with an operational domain;\nthey have problems to deal with new emerging tasks, such as concept shift in\ninput data that is used for learning. To tackle this challenge, we present\n\\textit{lifelong self-adaptation}: a novel approach to self-adaptation that\nenhances self-adaptive systems that use ML techniques with a lifelong ML layer.\nThe lifelong ML layer tracks the running system and its environment, associates\nthis knowledge with the current tasks, identifies new tasks based on\ndifferentiations, and updates the learning models of the self-adaptive system\naccordingly. We present a reusable architecture for lifelong self-adaptation\nand apply it to the case of concept drift caused by unforeseen changes of the\ninput data of a learning model that is used for decision-making in\nself-adaptation. We validate lifelong self-adaptation for two types of concept\ndrift using two cases.\n
