讲座题目：Latent Factorization of High-Dimensional and Incomplete Tensors

讲座时间：2024年4月2日15点

讲座地点：行政楼1楼多功能会议厅

讲座简介：Complex and temporal interactions among numerous nodes are frequently encountered in large-scale big data-related applications such as the recommender systems, social network service systems, and cryptocurrency network transaction systems. Such interactions data can be quantized into a step-N (N≥3) tensor whose most entries are unknown, i.e., a high-dimensional and incomplete (HDI) tensor. Despite its highly incompleteness, such an HDI tensor contain rich information regarding various desired patterns like the unknown interactions or undetected communities. To discover such patterns, this talk presents the latent factorization of tensors (LFT) models. An LFT model addresses the known data of the target HDI tensor in a data density-oriented way and establish highly efficient optimization algorithms for extracting desired latent features from it, thus implementing its representation learning accurately and efficiently. An LFT model has the great potential for industrial usage owing to its high efficiency in both computation and storage.

报告人简介：

罗辛，工学博士、博士后，西南大学二级教授、博士生导师，计算机与信息科学学院副院长，江西师范大学“正大讲座教授”。研究聚焦数据科学领域，在IEEE T. PAMI、IEEE T. KDE、IEEE T. NNLS等国际期刊和WWW、ICDM等国际会议上发表学术论文200余篇（含IEEE Transactions/Journal论文108篇、ESI高引论文27篇），累计影响因子超过1000，Web of Science统计引用超过4000次，谷歌学术统计引用超过9000次，H指数为58。先后主持国家级项目7项，省部级项目10余项，累积负责科研经费超过5000万元。获国家发明专利授权35项、实现27项授权专利的成果转化，累积产生经济效益超过1亿元。获国家万人计划青年拔尖人才、中国科学院百人计划、重庆市杰出青年基金等人才项目支持。获重庆市自然科学一等奖（2019/排名1）、重庆市科技进步一等奖（2018/排名2）、中国人工智能学会吴文俊人工智能科技进步一等奖（2018/排名3）等科技奖励。现任中国科技期刊卓越行动计划重点类期刊IEEE/CAA Journal of Automatica Sinica的副主编（Deputy Editor-in-Chief）、神经网络领域国际著名期刊IEEE Transactions on Neural Networks and Learning Systems的副编辑（Associate Editor），曾获IEEE/CAA Journal of Automatica Sinica的2020年度杰出副编辑奖。