The research introduces the Continuous Time Continuous Space Homeostatic Regulated Reinforcement Learning (CTCS-HRRL) framework. This framework extends the Homeostatic Regulated Reinforcement Learning (HRRL) framework to a continuous time-space environment. The authors design a model that mimics homeostatic mechanisms in a real-world biological agent using the Hamilton-Jacobian Bellman Equation, neural networks, and Reinforcement Learning. The model’s efficacy is demonstrated through a simulation-based experiment that shows the agent learning homeostatic behavior in a continuous time-space environment. The results suggest that the CTCS-HRRL framework is a promising tool for modeling animal dynamics and decision-making.
Publication date: 17 Jan 2024
Project Page: https://arxiv.org/abs/2401.08999
Paper: https://arxiv.org/pdf/2401.08999
