Continuous Modeling Perspective for Imaging Science

发布时间：2025-06-16 作者：浏览次数：

Speaker:	赵熙乐	DateTime:	2025年6月23日（周一）上午10:00--11:00
Brief Introduction to Speaker: 赵熙乐，电子科技大学教授、博导，入选四川省学术和技术带头人、四川省天府青城计划。第一/通讯在权威期刊SIAM 系列(SISC和SIIMS)和IEEE系列(TPAMI、TIP、TSP和TNNLS等)及顶会CVPR等发表研究工作。研究成果获四川省自然科学一等奖、四川省科技进步一等奖、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、四川省杰出青年科学基金项目、华为项目。		Place:	国交2号楼201
Abstract:To tackle inverse problems in imaging science, the regularizer, serving as an indispensable cornerstone in modeling, are usually introduced. In this talk, we will begin by reviewing the classical regularizers, including local regularizers, nonlocal regularizers, and global regularizers. We then 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 tiny neural networks. This paradigm allows us to readily deconstruct and reconstruct the classical regularizers, thus unleashing the potential of regularizers. Extensive experiments demonstrate the promising performance of the continuous modeling perspective.