作者: Shu-Hao Chang , Yu-Jen Chiou , Chun Yu , Chii-Wann Lin , Tzu-Chien Hsiao
DOI: 10.1007/978-3-540-89208-3_34
关键词: Training set 、 Multivariate analysis 、 Partial least squares regression 、 Regularization (mathematics) 、 Adaptability 、 Noise reduction 、 Artificial intelligence 、 Statistics 、 Computer science 、 Test data 、 Overfitting 、 Pattern recognition
摘要: In this paper, we develop a novel Partial Regularized Least Squares (PRLS) method which combined regularization algorithm with (PLS) analysis for noise reduction application. general, and PLS fall into an overfitting problem ill-posed condition. It means that some feature selections make the training data to have better adaptability model, but quality of prediction would be poorly compared testing information. We usually expected selected model should consistent predicted result between data.