Topological Data Anlysis (tda) based Machine Learning Models for Biomolecular Data Analysis

30.10.2020  23:50
主  讲  人  : Kelin Xia        博士

活动时间: 10月31日14时00分       

地            点  : Zoom

讲座内容:

In this talk, I will talk about topological data analysis (TDA) and itsapplication in biomolecular data analysis, in particular, drug design. I willbriefly introduce the TDA-based learning models, including simplicial complexgeneration, feature generation, and their combination with learning models.Further, I will discuss our recently-proposed persistent spectral based machinelearning (PerSpect ML) models. Persistent spectral models, including persistentspectral graph, persistent spectral simplicial complex, and persistent spectralhypergraph, are based on spectral graph theory, spectral simplicial complextheory, and spectral hypergraph theory, respectively. Different from allprevious spectral models, a filtration process is introduced to generatemultiscale spectral models. More specifically, from the filtration process, aseries of nested topological representations, i,e., graphs, simplicialcomplexes, and hypergraphs, can be systematically generated and their spectralinformation can be obtained. Persistent spectral variables are defined as thefunction of spectral variables over the filtration value. We test our models onthe most commonly-used databases, including PDBbind-2007, PDBbind-2013, andPDBbind-2016. Our results are better than all existing models, for all thesedatabases, as far as we know.

主讲人介绍:

Dr. Kelin Xia obtained his PhD degree from the ChineseAcademy of Sciences in Jan 2013. He was a visiting scholar in the Department ofMathematics, Michigan State University from Dec 2009-Dec 2012. From Jan 2013 toMay 2016, he worked as a visiting assistant professor at Michigan StateUniversity. He joined Nanyang Technological University at Jun 2016. His research workfocuses on using differential geometry, algebraic topology, and statisticallearning to study biomolecular structure, flexibility, dynamics, and functions.Recently, he has been interested in analysis of topological data (TDA), modelmachine learning / deep learning based on topology and their application indrug development.

发布时间:2020-10-30 15:55:57