A new concept learning approach applied in diagnostic systems

The authors first discuss some important issues that should be considered when designing a good learning system and then investigate a concept learning strategy called the MA (memory aggregate) approach. To improve the performance of MA and also to take the important issues into account, a novel version of MA which is more suitable for real world applications is then proposed. To evaluate the performance of the approach, two experiments of diagnosing brain tumors and diagnosing acute exanthemas were further conducted. According to the experimental results, it is concluded that the MA approach seems to have good performance in such diagnostic systems