报告专家:陈韵梅 佛罗里达大学 教授
报告时间:2023年02月17日9:00—10:00
报告地点:腾讯会议ID:288-481-779 密码:230217
正阳楼二楼会议室
报告人简介:
陈韵梅,佛罗里达大学Distinguished Professor、博士生导师,致力于数学、图像处理和机器学习等交叉学科的研究,研究领域包括医学图像分析中数学模型的建立与数值优化方法的发展,并对其中潜在的数学理论进行了深入的研究。曾获中国国家自然科学三等奖和教育部科技进步一等奖,获国际发明专利9项,主持国家级项目30余项,在Inventiones Mathematicae, SIAM Journal on Imaging Science, Pattern Recognition,IEEE Transactions on Biomedical Engineering等杂志上发表学术论文200余篇。陈韵梅教授被公认为偏微分方程与图像处理领域内的国际知名科学家,在国内外具有崇高的学术地位。
报告概要:
In this talk I will present our variational deep learning method for interpretable and generalizable MRI reconstruction to address the task specific and extremely data demanding problems in deep learning-based approaches.We propose a learnable variational modelin which the regularization function is parameterized by two sets ofparameters: a task-invariant set for common feature encoding and a task specific part to account for the variations in the heterogeneous data.The network architecture follows exactly a convergent learned optimization algorithm for solving the nonconvex and nonsmooth variational model.The network is trained by a bileve optimization algorithm to prevent overfitting and improve generalizability A series of experimental results indicate that the proposed method generalizes well to the reconstruction problems whose undersampling patterns and trajectories are not present during training.