Semi-Supervised Classification Based on Laplacia Support Vector Machine with Quasi-Linear Kernel
主講人：Jinglu Hu, Waseda University
時間：2024年5月29日 10:00-11:10
地點：國際學院306
主持人：金洲
主講人簡介（Short Bio）：
Prof. Jinglu Hu received an M.Sci degree in Electronic Engineering from Sun Yat-Sen University, Guangzhou, China in 1986, and a Ph.D degree in Computer Science and System Engineering from Kyushu Institute of Technology, lizuka, Japan in 1997. From 1986 to 1988, he worked as a Research Associate and from 1988 to 1993 a Lecturer at Sun Yat-Sen University. From 1997 to 2003, he worked as a Research Associate at Kyushu University. From 2003 to 2008, he worked as an Associate Professor and since April 2008, he has been a Professor at Graduate School of Information, Production and Systems of Waseda University. His research interests include Computational Intelligence and its applications to system modeling and identification, bioinformatics, time series prediction, financial analysis, data mining and so on. Dr. Hu has published more than 190 journal papers indexed by SCl or El and 150 prominent conference papers. Dr. Hu is a member Of IEEE, IEEJ, SICE and lEICE.
學術(shù)報告內(nèi)容摘要 （Abstract）：
In this talk, we first introduce a semi-supervised classification algorithm based on Laplacian support vector machine (SVM) with a quasi-linear kernel. The algorithm consists of two steps. In the first step, it composes a quasi-linear kernel through implementing a piecewise linear model based on neural networks. In the second step, the piecewise linear model is further optimized by a Laplacian SVM algorithm using the quasi-linear kernel function. In both steps, it effectively leverages a small amount of labeled and a large amount of unlabeled data so as to achieve good classification performance. Then we introduce an application of the algorithm to construct a semi-supervised classifier for parasite image classification, by including a semi-supervised feature extractor based on deep CNN using contrastive learning.