We use gene regulatory networks (GRNs) to modulate the parameters of the SARSA reinforcement learning algorithm, demonstrating that GRNs can be optimized to enhance learning and generalize across multiple tasks without specific problem knowledge.
Jul 11, 2015
This study uses neuromodulatory gene regulatory networks to improve reinforcement learning agents in solving the robot coverage control problem, demonstrating better generalization and learning capabilities compared to fixed parameter settings.
Aug 4, 2013
This paper presents a system that uses local diffusion and reinforcement learning to predict an entire input space based on partial observation, addressing bandwidth constraints in real-time input. It optimizes for multi-dimensional input spaces and maintains the ability to react to rare events.
Sep 24, 2007