Approximation of an analog diffusion network with applications to image estimation (Q5925728)

The authors develop a numerical procedure to approximate Wong's diffusion machine, which is a stochastic neural network. They use a discretization of the stochastic differential equations and show that an appropriately scaled sequence of the discrete iterates converges to that of the analog diffusion machine.They also provide some examples of the application of diffusion networks for image segmentation and restoration.