Abstract
The CBR (case based reasoning) is a problem solving technique following different strategy compared to the major approaches of the artificial intelligence. It develops remedies to certain problem based on the pre-existing solutions of similar nature. So the problem using the CBR is handled by retrieving and reusing the similar previously solved problems and available solutions respectively. This makes the process functioning alike based on the human activities is instinctively attractive and more beneficial compared to the Conventional_AI as begins to reason out the possible solutions form the shallow base. The CBR due to the exceeding performance are popular among a wide range of applications such as the weather fore casting, medical and engineering diagnosis, aerospace etc. Identification or sorting out or classification take a significant role in cases that is the training examples retrieval as the perfect identification results in perfect case retrieval, this further enables the case based reasoning to arrive to at a perfect remedy for the problem. The retrieval of cases are mostly based on the similarity and utilizes the KNN (K-Nearest Neighbor). The proposed method in the paper integrates the multilayer perceptron with the fuzzy nearest neighbor (MLP-NFF) system with the help of WEKA to deliver a perfect classification to make the CBR-retrieval efficient. The evaluation of the proposed method and its comparison with the KNN is done using the standard data set obtained from the medical field.
Publisher
Inventive Research Organization