Rule Induction Based on Logic Synthesis Methods

Abstract

Application of logic synthesis methods for data mining is discussed. The key concept is to apply the Boolean function complement algorithm for rule induction. The presented results of experiments with large medical databases indicate that the proposed approach significantly improves the efficiency of the rule induction procedure. Compared with the earlier presented, commonly used algorithm, the average rule accuracy has increased by 10% and the rule coverage by 15%, ultimately reaching 81.5% and 97.0%, respectively. © Springer International Publishing Switzerland 2015.