Constrained reinforcement learning for personalization in highly regulated domains

Adaptivity and personalization can be of huge benefit in highly regulated domains such as healthcare and finance. Strong guarantees on safety of behaviour is a prerequisite for adoption of systems in these domains. This project is aimed at (a) bridging the gap between regulators’ and agent’s behaviour representation and (b) reinforcement learning under the resulting constraints aimed at the application of an Adaptive Personal Assistant.
This project is a collaboration with the Knowledge Representation and Reasoning group and the Core Banking University at ING

Partner

Team

Floris den Hengst
PhD Student

Floris den Hengst

Dr. Mark Hoogendoorn
Assistent Professor

Dr. Mark Hoogendoorn

Prof. dr. Guszti Eiben
Head of the Group

Prof. dr. Guszti Eiben