BLOG AND NEWS
Reinforcement learning for bioprocesses - a study of reinforcement learning algorithms for continuous bioprocessing
The cited GIT project presents “reinforcement learning” techniques for the optimization of continuous bioprocesses that have become popular in recent years due to their high productivity over longer process times.
The Bioprocess model considers a simple fermentation with an improved concentration of the molecule of interest allocated by back-allocation from the terminal state of the fermentation and an associated yield of some objective function.
Three Reinforcement Learning models are used in the GIT project:
– Monte Carlo
– SARSA
– Q learning
The ExpDesign.py file calls all other “.py” scripts and functions. The current file is configured to train and validate only the Monte Carlo model can be adjusted by changing the import in ExpDesign.py.