Machine Learning in Granular Computing

摘要

Main purpose of the Granular Computing (GrC) is to find a novel way to acquire knowledge for huge orderless very high dimensional perception information. Obviously, such kind Granular Computing (GrC) has close relationship with machine learning. In this paper, we try to study the machine learning under the point of view of Granular Computing (GrC). Granular Computing (GrC) should contain two parts: (1) dimensional reduction, and (2) information transformation. We proved that although there are tremendous algorithms for dimensional reduction, their ability can't transcend the old fashion wavelet kind nested layered granular computing. To change a high dimensional complex distribution domain to a low dimensional and simple domain is the task of information transformation. We proved that such kind mapping can be achieved as a granular computing by solving a quadric optimization problem.