Add huber to ADMM - Githubissues

benchopt / benchmark_tv_1d

TV Denoising in 1D

2 stars 7 forks source link

Add huber to ADMM #42

Closed EnLAI111 closed 2 years ago

EnLAI111 commented 2 years ago

I try to add huber loss to ADMM, however, scipy.optimize.minimize can't succeed every time. And the convergence is too slow.

EnLAI111 commented 2 years ago

It works for convolution, but not for multiplication. But I can't see if there is something different between the convoluted case and multiplicative case with Huber.

EnLAI111 commented 2 years ago

It converges!!