Provably convergent plug-and-play splitting methods for nonconvex learning problems with applications
发布时间：2024-04-17 点击次数：
标题：Provably convergent plug-and-play splitting methods for nonconvex learning problems with applications
报告时间：2024年04月22日（星期一）09:00-10:00
报告地点：人民大街校区数学与统计学院105教室
主讲人：吴中明
主办单位：数学与统计学院
报告内容简介：
In this talk, I will introduce several splitting methods for nonconvex optimization problems, and then combine them with extrapolation and Plug-and-Play (PnP) prior. Specifically, we investigate the convergence properties and applications of the three-operator splitting method, also known as Davis-Yin splitting (DYS) method, integrated with extrapolation and Plug-and-Play (PnP) denoiser within a nonconvex framework. Our approach provides an algorithmic framework that encompasses both extrapolated forward-backward splitting and extrapolated Douglas-Rachford splitting methods. To establish the convergence of the proposed method, we rigorously analyze its behavior based on the Kurdyka-？ojasiewicz property, subject to some tight parameter conditions. Moreover, we introduce two extrapolated PnP-DYS methods with convergence guarantee, where the traditional regularization prior is replaced by a gradient step-based denoiser. Finally, we conduct extensive experiments on image deblurring and image super-resolution problems, where our results showcase the advantage of the extrapolation strategy and the superior performance of the learning-based model that incorporates the PnP denoiser in terms of achieving high-quality recovery images.
主讲人简介：
吴中明，南京信息工程大学副教授，香港中文大学博士后，东南大学博士，新加坡国立大学访问学者。研究方向为最优化理论、方法及其应用。在SIAM Journal on Imaging Sciences, IEEE Transactions on Signal Processing, Computational Optimization and Applications, Journal of Global Optimization, Mathematics of Computation, Annals of Operations Research等期刊发表/录用论文三十余篇。入选南京信息工程大学“青年科技之星”，江苏省“双创博士”，人社部博管办“香江学者计划”。担任中国运筹学会宣传工作委员会委员，中国运筹学会数学规划分会青年理事，江苏省运筹学会理事、副秘书长。主持国家自然科学青年基金项目，中国博士后面上资助项目，江苏省科技智库青年项目。