A Mulitscale Attentional Framework for Relaxation Neural Networks

We investigate the optimization of neural networks governed by general objective functions. Practical formulations of such objectives are notoriously difficult to solve; a common problem is the poor local extrema that result by any of the applied methods. In this paper, a novel framework is introduced for the solution oflargescale optimization problems. It assumes little about the objective function and can be applied to general nonlinear, non-convex functions; objectives in thousand of variables are thus efficiently minimized by a combination of techniques deterministic annealing , multiscale optimization, attention mechanisms and trust region optimization methods.