ADMM-NeuralNetworks
Training Neural Networks Without Gradients: An ADMM Approach (Matlab, Tensorflow1.6-Python2.7/3.5 with eager execution and matplotlib) Note: You cannot use this code for any assignment or any profit product.
This code was implemented follow below paper; however, I used Mean Square Error loss instead of Binary Hinge loss
Taylor, Gavin, et al. "Training neural networks without gradients: A scalable admm approach." International Conference on Machine Learning. 2016.
Only support for GPU?
I only implemented for GPU version and tested on MNIST data.
How to run?
Run main.m/main.py