HDLSS 자료에 대한 머신러닝 적용 - 암 진단 자료 중심으로

Abstract

HDLSS (high dimension low sample size) refers to high-dimensional but small data. HDLSS data appear frequently in genetic data. In this paper, we intend to examine the data for cancer diagnosis to which HDLSS data belongs. In this paper, we compare the performance of several machine learning classifiers on HDLSS data. In addition, the loss function is visualized. And multicollinearity and the influential points will be considered in HDLSS data. We look at the performance results of deep learning considering multicollinearity and influence points.