Finding the good experiment to reuse from the case memory is the key of success in Case Based Reasoning (CBR). The paper presents a novel associative memory model to perform this task. The algorithm is founded on a Particle Swarm Optimization (PSO) approach to compute the neighborhood of a new problem. Then, direct access to the cases in the neighborhood is performed. The model was experimented on the Adult dataset, acquired from the University of California at Irvine Machine Learning Repository and compared to flat memory model for performance. The obtained results are very promising.
展开▼