Editing Reinforcement Learning and Forgetting in Knowledge-Based Configuration

Title: Reinforcement Learning and Forgetting in Knowledge-Based Configuration

Abstract: This research article proposes a heuristic called Relevant Knowledge First (RKF) for making decisions in configuration processes. The RKF uses a relevance function that consists of two components: one based on reinforcement learning and the other based on forgetting (fading). The relevance of an object increases with its successful use and decreases with age when it is not used. This method aims to speed up the configuration process and improve the quality of solutions relative to the reward value given by users.

Main Research Question: How can we improve the efficiency and quality of knowledge-based configuration processes by developing a heuristic that considers the relevance of objects based on reinforcement learning and forgetting?

Methodology: The researchers developed the RKF heuristic for making decisions in configuration processes. The relevance function has two components: one based on reinforcement learning and the other based on forgetting (fading). The relevance of an object increases with its successful use and decreases with age when it is not used.

Results: The researchers found that the RKF heuristic can speed up the configuration process and improve the quality of solutions relative to the reward value given by users. This was achieved by considering the relevance of objects based on their successful use and age.

Implications: The RKF heuristic has the potential to improve the performance and quality of knowledge-based configuration processes. By using a relevance function that considers both reinforcement learning and forgetting, the heuristic can help to make better decisions and avoid unnecessary revision, thereby accelerating the configuration process.

Link to Article: https://arxiv.org/abs/0109034v1
Authors: 
arXiv ID: 0109034v1

[[Category:Computer Science]]
[[Category:Based]]
[[Category:Configuration]]
[[Category:Relevance]]
[[Category:Heuristic]]
[[Category:Reinforcement]]