Skip to content

An implementation of the Binarized Neural Networks

Notifications You must be signed in to change notification settings

grassking100/bnn

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

4 Commits
 
 
 
 
 
 
 
 

Repository files navigation

An implementation of the Binarized Neural Networks

Dataset:

MNIST dataset

Method:

There were two kinds of models named binarized neural networks (1-bit) and full-precision model (16-bits). Each model has three building blocks.

Results:

  1. Loss curve

loss

  1. Macro F1 curve

macro_f1

  1. Test confusion matrix of full-precision model

full_confusion

  1. Test confusion matrix of binarized neural networks

binary_confusion

  1. Test result
Model Epoch Macro F1 Loss
Full-precision model 33 0.991 0.0382
Binarized neural networks 35 0.978 0.2589
  1. Visualization of some kernels of the models
Model Kernel 0 Kernel 1 Kernel 2
Full-precision model f0 f1 f2
Binarized neural networks b0 b1 b2
  1. Weights
  • Full-precision model: link

  • Binarized neural networks: link

References:

Simons, Taylor, and Dah-Jye Lee. "A review of binarized neural networks." Electronics 8.6 (2019): 661.

About

An implementation of the Binarized Neural Networks

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages