Conquering Uncertainty in Multiple-Valued Logic Design

摘要

In modern science, significant advances are typically made atcross-roads of disciplines. Thus, many optimization problems inMultiple-valued Logic Design have been successfullyapproached using ideas and techniques from ArtificialIntelligence. In particular, improvements in multiple-valuedlogic design have been made by exploiting information/uncertaintymeasures. In this paper, we review well-known information measuresin the multiple-valued domain and consider some methods of findinginformation measures for completely or incompletely specifiedfunctions with multiple-valued and continuous attributes. In thisrespect, the paper addresses the problem known as discretizationand introduces a method of finding an optimal representation ofcontinuous data in the multiple-valued domain. We also propose atechnique for efficient calculation of different informationmeasures using Multiple-valued Decision Diagrams. As oneapplication of our technique, we outline an approach tosynthesizing digital circuits derived from decision diagrams thatcan yield to reduction in power dissipation. The paper also showsthe impact in several important areas of multiple-valued systemdesign including (i) fuzzy logic, (ii) quantum computing systems,and (iii) data mining.