A Learning Algorithm based on High School Teaching Wisdom

摘要

A learning algorithm based on primary school teaching and learning ispresented. The methodology is to continuously evaluate a student and to givethem training on the examples for which they repeatedly fail, until, they cancorrectly answer all types of questions. This incremental learning procedureproduces better learning curves by demanding the student to optimally dedicatetheir learning time on the failed examples. When used in machine learning, thealgorithm is found to train a machine on a data with maximum variance in thefeature space so that the generalization ability of the network improves. Thealgorithm has interesting applications in data mining, model evaluations andrare objects discovery.