报告人：赵熙乐（电子科技大学）

时间：2024年08月29日 14:30

地址：理科楼LA103

摘要：The regularizer, which incorporates prior knowledge of images, is cornerstone of inverse imaging problem modelling. In this talk, we will first review the classical regularizers, including total variation regularizer, low-rank regularizer, and nonlocal regularizer. Then we will discuss the limitations of classical hand-crafted regularizers (e.g., expressive capability, applicability, and flexibility). To address the above limitations of classical regularizers, we suggest a unified Continuous Modeling Perspective for imaging science, which continuously represents discrete data by elegantly leveraging a deep neural network. This paradigm allows us to deconstruct and reconstruct the classical regularizers readily. Extensive experiments demonstrate the promising performance of the continuous modeling perspective.

简介：赵熙乐，电子科技大学教授、博导，中国工业与应用数学学会副秘书长，入选电子科技大学百人计划和四川省学术和技术带头人后备人选。撰写Elsevier出版社和科学出版社出版的学术专著章节2章，第一/通讯在权威SIAM 系列期刊(SIIMS和SISC)和IEEE系列期刊(TIP、TNNLS、TCYB、TCVST、TCI和TGRS)、ISPRS及计算机学会A类会议CVPR和AAAI等发表研究工作。研究成果获四川省科技进步一等奖两项(自然科学类、科技进步类)、第一、二届川渝科技学术大会优秀论文一等奖，计算数学会青年优秀论文竞赛二等奖。主持国自然面上项目(2项)、国自然青年项目、华为项目。

邀请人：李寒宇

欢迎广大师生积极参与！