A Cognitive Approach to Forgetting in Case-Based Reasoning by Combining Cased-Based Reasoning and Artificial Neural Networks

Abstract Forgetting in knowledge-based systems allows the systems to adapt in dynamic environments by removing irrelevant information. Traditional deletion strategies used in case-based reasoning systems are good at improving system performance while preserving competence. However, these systems have been proven to work rigidly when applied to assistive technologies because they do not possess the same forgetting mechanisms as humans do. In this project, I plan to add an artificial neural network into the case-based reasoning, so that the network will act as the neocortex and the case base will serve as the hippocampus in the complementary learnin systems. The addition of artificial neural network allows for extraction of general information from the case base, so that the case base can delete specific cases and learn new information. This type of forgetting in the case base is similar to systems consolidation in the complementary learning system, where memory is transferred out of the hippocampus to the neocortex. The main goal of this project is to take on case-based reasoning from a cognitive neuroscience perspective. In doing so, the objective is to see if artificial neural network-based case-based reasoning system will be able to capture mechanisms of the human mind, while improving upon traditional case based reasoning systems in terms of both performance and competence.