RECALL of multilayer perceptron

摘要: A method, called RECALL, which provides a way for directly perceiving multilayer perceptron's learning is discussed. After the process of networks has been accomplished, this method used to recall what network learned, or underlying rule it had. The result will reflect network's internal character because multi layer perceptron discernment mainly corresponds it. By was found that in with one hidden relevant information representation significantly reflected upon result. different desired value would attract attention places. With believed more reasonable apply implies relationship among training samples. This RECALL can be analyze other problems and some implication similarity between connectionist model human beings. >