报告题目:Selection dynamics for Deep Neural Networks
报 告 人:刘海亮 爱荷华州立大学教授
报告时间:2023年6月21日周三下午4:00
报告地点:S3-502
摘要:We will present a partial differential equation framework for deep residual neural networks and for the associated learning problem. This is done by carrying out the continuum limits of neural networks with respect to width and depth. We study the well-posedness of the forward problem, and establish several optimal conditions for the inverse deep learning problem.This talk concerns several mathematical aspects of deep learning and the use of optimal control tools in solving the learning problem. This presentation is based on a joint work with Peter Markowich (KAUST).
个人简介:
刘海亮,美国爱荷华州立大学数学与计算科学系教授。河南师范大学数学学士学位,清华大学数学硕士学位,中国科学院数学博士学位。主要研究方向为偏微分方程分析,发展解决偏微分方程问题的高阶数值算法及应用等。曾获得多项荣誉和奖项,包括德国洪堡学者,应用数学首席教授(Holl Chair)。最近的工作集中在研究数据驱动的深度学习问题。
上一条:学术报告会:Globally smooth solutions to time-like extremal surface equations with large data 下一条:学术报告会:Liouville type theorems and a priori estimates
【关闭】