Utility Based State Learning for Controlling Partially Observable Gene Regulatory Networks
The external control of Gene Regulatory Networks (GRNs) has received much attention in recent years. In this paper, we propose a novel algorithm for controlling partially observable GRNs coupling utility based state learning methods and Batch Mode Reinforcement Learning (Batch RL) methods. Our aim is to construct an infinite horizon Partially Observable Markov Decision Process (POMDP) model directly from the available gene expression data. Results show that our novel POMDP model produces better control policies than the available solutions and requires less time.