Facilitating Safe Adaptation of Interactive Agents using Interactive Reinforcement Learning
Konstantinos Tsiakas · 2016
In this paper, we propose a learning framework for the adaptation of an interactive agent to a new user. We focus on applications where safety and personalization are essential, as Rehabilitation Systems and Robot Assisted Therapy. We argue that interactive learning methods can be utilised and combined into the Reinforcement Learning framework, aiming at a safe and tailored interaction.