An inexact restoration-nonsmooth algorithm with variable accuracy for stochastic nonsmooth convex optimization problems in machine learning and stochastic linear complementarity problems

We study unconstrained optimization problems with nonsmooth and convex objective function in the form of a mathematical expectation. The proposed method approximates expected sample average using Inexact Restoration-based adapted sizes. size is chosen an adaptive manner based on Restoration. algorithm uses line search assumes descent directions respect to current approximate function. prove a.s. convergence under standard assumptions. Numerical results for two types problems, machine learning loss training classifiers stochastic linear complementarity efficiency scheme.