Abstract—A novel and efficient method for the Classifier Integration Model(CIM) by adopting the concept of confusion table is extensively studied and reported in this paper. The method considers not only the diagonal elements of each confusion table but also the non-diagonal elements obtained from the existing confusion tables for local classifiers in CIM for more accurate classification performance. The CIM with Confusion Table (CIM-CT) method is applied to two different data sets, Iris data set and an audio signal data set for evaluation of the CNN-CT scheme. The experimental results show that the CIM-CT method outperforms a conventional classifier and the classifier with utilizing only the diagonal elements of confusion tables in CIM in terms of classification accuracy.
Index Terms—Classifier, features, confusion table, machine learning.
Miso Jang was with Myongji University, Yongin, 17058, Rep. of Korea. She is now with the Mindin Tech., Seoul, 05510 Rep. of Korea (e-mail: firstname.lastname@example.org).
Dong-Chul Park is with the Department of Electronics Engineering, Myongji University, Gyeonggi-do, Yongin, 17058 Rep. of Korea (e-mail: email@example.com).
Cite: Miso Jang and Dong-Chul Park, "Application of Classifier Integration Model with Confusion Table to Audio Data Classification," International Journal of Machine Learning and Computing vol. 9, no. 3, pp. 368-373, 2019.