https://github.com/wangg12/irls_using_dl_frameworks

IRLS(Iterative re-weighted least square) for Logistic Regression, implemented using tensorflow2.0/pytorch
https://github.com/wangg12/irls_using_dl_frameworks

irls logistic-regression python3 pytorch tensorflow tensorflow-2

Last synced: about 1 year ago
JSON representation

IRLS(Iterative re-weighted least square) for Logistic Regression, implemented using tensorflow2.0/pytorch

• Host: GitHub
• URL: https://github.com/wangg12/irls_using_dl_frameworks
• Owner: wangg12
• License: apache-2.0
• Created: 2017-03-24T12:16:26.000Z (over 9 years ago)
• Default Branch: master
• Last Pushed: 2025-04-02T04:14:35.000Z (about 1 year ago)
• Last Synced: 2025-04-12T17:09:48.340Z (about 1 year ago)
• Topics: irls, logistic-regression, python3, pytorch, tensorflow, tensorflow-2
• Language: Python
• Homepage:
• Size: 645 KB
• Stars: 13
• Watchers: 3
• Forks: 6
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

          
# IRLS using DL frameworks

* IRLS(Iterative re-weighted least square) for Logistic Regression, implemented using

  * tensorflow < 2.0

  * tensorflow2.0

  * pytorch

  * paddlepaddle

  * megengine

* Note that IRLS is a second order optimization problem, which is equivalent to Newton's method.

* We show that these DL frameworks can do general matrix based algorithms, and can be accelerated by the power of gpu.

* In this implementation, we use svd to solve pseudo inverse of singular matrices.