HUMAN-CENTERED PERSPECTIVES IN INTERACTIVE MACHINE LEARNING FOR ADVANCING AMBIENT INTELLIGENCE

Abstract

Abstract-As the vision of Ambient Intelligence (AmI) becomes more feasible, the challenge of designing effective
and usable human-machine interaction in this context becomes increasingly important. Interactive Machine Learning
(IML) offers a set of techniques and tools to involve end-users in the machine learning process, making it possible to
build more trustworthy and adaptable ambient systems. In this paper, our focus is on exploring ap proaches to
effectively integrate and assist human users within ML-based AmI systems. Through a survey of key IML-related
contributions, we identify principles for designing effective human-AI interaction in AmI applications. We apply
them to the case of Op portunistic Composition, which is an approach to achieve AmI, to enhance collaboration
between humans and Artificial Intelligence. Our study highlights the need for user-centered and context-aware
design, and provides insights into the challenges and opportunities of integrating IML techniques into AmI systems