A practical neural network for handwritten character recognitionbuilt by dynamics-based active learning and self-organization offeedback mechanism

摘要

A feedback mechanism which enhances similarity between recognitionby a human and that by an artificial neural network we have proposed isintroduced. A human may turn his attention to a particular portion ofthe object in order to classify it successfully. It is regarded as afeedback from the first rough classification to the final one. Theproposed feedback mechanism is designed to simulate such attention. Therecognition rates by a neural network with the feedback mechanism exceedthose by a former neural network