JSAI2019

Presentation information

General Session

General Session » [GS] J-2 Machine learning

[4I3-J-2] Machine learning: analysis and buliding of basic models

Fri. Jun 7, 2019 2:00 PM - 3:20 PM Room I (306+307 Small meeting rooms)

Chair:Takuya Hiraoka Reviewer:Shohei Higashiyama

2:40 PM - 3:00 PM

[4I3-J-2-03] Comparison of Rotation Criteria of Factor Analysis and Independent Component Analysis

〇Yuto Imamura1, Takahiro Nishigaki1, Takashi Onoda1 (1. Aoyama Gakuin University)

Keywords:Factor Analysis, Independent Component Analysis, Rotation Criteria, Kurtosis Criterion, Crawford and Ferguson Criterion

The purpose of this research is to clarify the difference of the characteristics of independent component analysis and factor analysis.

We focus on the rotation criterion of these two methods this paper.

we clarified the relationship between Crawford and Ferguson criterion used in factor analysis and kurtosis criterion used in independent component analysis.

Moreover, we found that the characteristics of the kurtosis criterion that emphasizes the simplification of rows of a factor loading matrix because the relationship between Crawford and Ferguson criterion and kurtosis criterion.

We can make the same rotation as the kurtosis criterion by changing the weights of Crawford and Ferguson criterion.

Experimental using sample data showed the relationship between this kurtosis criterion and Crawford and Ferguson criterion and the characteristics of kurtosis.