Interval-state cellular automata and their applications to image segmentation

We present a new type of optimization algorithms, adapted for neural network training. These algorithms are based upon sequential operator splitting technique for some associated dynamical systems. Furthermore, we investigate through numerical simulations the empirical rate of convergence of these iterative schemes toward a local minimum of the loss function, with some suitable choices of the underlying hyper parameters. We validate the convergence of these optimizers using the results of the accuracy and of the loss function on the MNIST,MNIST-Fashion and CIFAR 10 classification datasets.

I.D. Voiculescu
(Oxford, MPLS, Computer Science)

I. Boros
(Tiberiu Popoviciu Institute of Numerical Analysis, Romanian Academy, Cluj-Napoca, Romania)

N. Popovici

L. Diosan

A. Andreica

L. Dioşan, A. Andreica, I. Boros, I. Voiculescu, N. Popovici,  Interval-state cellular automata and their applications to image segmentation, SAGE Publications, 2017,
https://ora.ox.ac.uk/objects/uuid:eacaf9ce-99c2-49dd-8e9e-faced821233a

google scholar link