Data Science| The Pruning Technique of Neural Networks Related to Compressed Sensing

报告人：夏羽博士（杭州师范大学）

报告题目：The Pruning Technique of Neural Networks Related to Compressed Sensing

时间与地点：2019年10月28日下午3:30-5:30,玉泉校区教十一417
报告摘要：We discuss the Net-Trim, which is a technique simplifies a trained neural network. The method is a convex post-processing module, which prunes a trained network layer by layer, while preserving the internal responses. The initial and retrained models before and after Net-Trim application is consistent and the number of training samples needed to discover a network can be expressed using a certain number of nonzero terms. Specifically, if there is a set of weights that uses at most s terms that can re-create the layer outputs from the layer inputs, these weights can be found from O(slog N/s) samples, where N is the input size. The related theoretical results are similar to those for sparse regression using the Lasso and the tools are related to concentration of measure and convex analysis.

联系人：李松教授（songli@zju.edu.cn）