A theoretical Q-learning temporary security repair

This research summarizes the first attempt to incorporate Q-learning algorithm in software security. The Q-learning method is embedded as part of the software itself to provide a security mechanism that has ability to learn by itself to develop a temporary repair mechanism. The results of the experiment express that given the right parameters and the right setting the Q-learning approach rapidly learns to block all malicious actions. Data analysis on the Q-values produced by the software can provide security diagnostic as well. A larger scale experiment with extended parameter testing is expected to be seen in the future work.