北京雁栖湖应用数学研究院

丘成栋

陶飏天择

2023年06月20日 15:00 至 15:30

数学系理科楼A-203

In this report, we will introduce how to combine optimization with neural networks, more precisely, by embedding convex optimization process as a network layer into neural networks, enabling gradient propagation. This work is based on OptNet: Differentiable Optimization as a Layer in Neural Networks and Differentiable Convex Optimization Layers.