演讲摘要：The performance of most clustering methods hinges on the used pairwise affinity, which is notoriously known for its vulnerability of noise contamination or the imbalance in samples or features. This talk proposes to use the high order information among pairs to boost the clustering performance. We defined two ways to construct such high order information, which is denoted as the tensor similarity. We proved that the former equals the Kronecker product of pairwise similarity matrices, and the latter provide complementary information for which the pairwise similarity missed. Hence, the high order similarity matrix obtained from the second one serves as a robust complement for the pairwise similarity. Integrating it with the popular pairwise similarity (IPS2) could efficiently boost the clustering performance. Extensive experiments demonstrated that the proposed IPS2 significantly outperformed previous similarity-based methods on real-world datasets, and it was capable of handling the clustering task over under-sampled and noisy datasets.
讲者简介：华南理工大学计算机科学与工程学院教授，博士生导师，京都大学客座教授。2016年科技部重点领域创新团队核心成员，2016广东省计算智能与网络空间信息重点实验室核心成员，全国系统生物学专业委员会委员，生物信息学与人工生命专业委员会委员、常委委员，计算机协会生物信息学专业委员会委员、常委委员，CCF高级会员。2012年9月至今在华南理工大学任教，2016年9月破格晋升博士生导师，同年破格晋升教授。长期参与生物医学图像和生物信息处理方面的研究工作，在医学图像分析与理解、生物信息分析、多源数据融合、模式识别和数据挖掘等领域积累了丰富的研究经验。他在相关领域发表论文70余篇，包括Bioinformatics、Brief in Bioinformatics, Neuroimage、Medical Image Analysis, IEEE Trans Image Processing、IEEE Trans. Biomedical Engineering等。是国际会议ISB 2019/2018/2017/2016/2015/2014，ISBRA 2018/2017/2016, ICIC 2019/2018/2017/2016, BESC 2018, CBC 2018/2017/2016, CIBB2015, GIW 等会议的PC。国际会议ICBBB2020 的PC Chair。
第五届中国计算机学会生物信息学会议
分会场邀请报告
京公网安备11010802032778号