G-2022-27
A stochastic proximal method for nonsmooth regularized finite sum optimization
, , and BibTeX reference
We consider the problem of training a deep neural network with nonsmooth regularization to retrieve a sparse and efficient sub-structure. Our regularizer is only assumed to be lower semi-continuous and prox-bounded. We combine an adaptive quadratic regularization approach with proximal stochastic gradient principles to derive a new solver, called SR2, whose convergence
and worst-case complexity are established without knowledge or approximation of the gradient's Lipschitz constant. We formulate a stopping criteria that ensures an appropriate first-order stationarity measure converges to zero under certain conditions.
We establish a worst-case iteration complexity of O(ϵ−2)
that matches those of related methods like ProxGEN, where the learning rate is assumed to be related to the Lipschitz constant.
Our experiments on network instances trained on CIFAR-10 and CIFAR-100 with ℓ1
and ℓ0
regularizations show that SR2 consistently achieves higher sparsity and accuracy than related methods such as ProxGEN and ProxSGD.
Published June 2022 , 17 pages
Research Axis
Research application
Document
G2227.pdf (10 MB)