报告题目:A newly constructed kernel in SVM and its applications
报告时间:2021年5月29日,星期六,上午8:30-11:30(北京时间)
腾讯会议:320 609 873 ,密码:0529
报告人: 姜昊副教授,中国人民大学
报告摘要:
Support vector machines (SVMs) are learning algorithms based on structural risk minimization, compared to traditional learning algorithms that are developed based on empirical risk minimization. One of the key attributes in SVMs lies in the construction of kernels, which are used to model the relationships among the training data. In this talk, we will present our newly constructed kernel: Hadamard kernel, for modeling the nonlinear relationship between high dimensional gene expression data. We will provide theoretical proof for the feasibility of Hadamard kernel in SVMs, and also give some applications in omics data mining problems.
报告人简介:
姜昊,现就职于中国人民大学数学学院,副教授,硕士生导师。博士毕业于香港大学,研究方向是计算生物信息学、机器学习与数据挖掘,在相关领域内的国际优秀期刊发表论文数篇,包括Bioinformatics,Applied Mathematical Modelling等,现研究基于单细胞的异质性分析,包括数学模型,计算方法等。受邀访问香港大学、日本京都大学等从事生物信息方面的合作研究。目前担任多个国际期刊的评审,包括Nucleic Acids Research,Journal of Theoretical Biology,IEEE/ACM Transactions on Computational Biology and Bioinformatics等。完成一项国家自然科学基金项目,目前主持一项国家自然科学基金,同时作为学术骨干参与国家自然科学基金重大研究计划集成项目。