arXiv 2307.08982
Neural Network Pruning as Spectrum Preserving Process
By Shibo Yao, Dantong Yu, et al.
Published 2023-07-18
Mindmap
Browse the paper's core ideas, clusters, and relationships in a structured outline.
Neural networks have achieved remarkable performance in various application domains. Nevertheless, a large number of weights in pre-trained deep neural networks prohibit them from being deployed on smartphones and embedded systems. It is highly desirable to obtain lightweight versions of neural networks for inference in edge devices. Many cost-effective approaches were proposed to prune dense and convolutional layer…