Https://arxiv.org/pd F/1911.13299.pdf

1911.13299.pdf
Preview of https://arxiv.org/pd f/1911.13299.pdf
🔗 Source: arxiv.org
📊 Size: 1.01 MB
📄 Pages: 13 pages
⬇️ Downloads: 116

Summary

Researchers demonstrate that randomly weighted neural networks contain subnetworks that can achieve impressive performance without modifying the weight values. They propose the edge-popup algorithm to find these subnetworks, which optimizes the scores of the weights via SGD without changing their values. Experiments on CIFAR-10 and ImageNet show that untrained subnetworks can perform as well as dense networks with learned weights, with a subnetwork of a randomly weighted Wide ResNet-50 matching the performance of a trained ResNet-34.

Description

Researchers demonstrate that randomly weighted neural networks contain subnetworks that can achieve impressive performance without modifying the weight values.

Technical Information

  • File Format: PDF
  • File Size: 1.01 MB
  • Pages: 13
  • Language: EN
  • Total Downloads: 116
  • Last Updated: 1 week ago

Document Overview

This PDF document about https://arxiv.org/pd f/1911.13299.pdf provides comprehensive information and guidance. Whether you're a beginner or advanced user, this resource offers valuable insights into https://arxiv.org/pd f/1911.13299.pdf.

Related Topics

If you're interested in https://arxiv.org/pd f/1911.13299.pdf, you might also want to explore:

Download https://arxiv.org/pd f/1911.13299.pdf eBooks for free and learn more about https://arxiv.org/pd f/1911.13299.pdf. These books contain exercises and tutorials to improve your practical skills, at all levels!

Not satisfied with this document? We have related documents to https://arxiv.org/pd f/1911.13299.pdf, try searching with similar keywords: https://arxiv.org/pd f/1911.13299.pdf, https://arxiv.org/ft p/arxiv/papers/1807/ 1807.03976.pdf, http://arxiv.org/ftp /arxiv/papers/1106/1 106.6178.pdf, https://arxiv.org/pd f/2306.12001.pdf, https://arxiv.org/pd f/2012.09696.pdf, https://arxiv.org/pd f/1609.03552v2.pdf, https://arxiv.org/pd f/1811.07073.pdf, https://arxiv.org/pd f/2303.11045.pdf

You can download PDF versions of the user's guide, manuals and ebooks about https://arxiv.org/pd f/1911.13299.pdf, you can also find and download for free A free online manual (notices) with beginner and intermediate, Downloads Documentation, You can download PDF files (or DOC and PPT) about https://arxiv.org/pd f/1911.13299.pdf for free, but please respect copyrighted ebooks.