Applications of Emergent Computation in Reaction-Diffusion CNNs for Image Processing

Abstract

The possibility to exploit emergent computation in a naturally inspired complex network, namely the reactiondiffusion cellular nonlinear network (RD-CNN), is investigated. The particular application under focus is image processing. It is shown that by implementing a simplified discrete-time model and by using the local activity theory to locate potentially useful regions in the huge parameter space, many useful image processing tasks may be performed in reasonable execution time. Such tasks may include but are not limited to: feature extraction, image enhancement, noise removal, pattern formation, etc. A framework is provided for a systematic design allowing the identification of useful genes (sets of parameters) associated with meaningful image processing tasks.

A Low Cost High Performance Computing Platform for Cellular Nonlinear Networks Using Python for CUDA

“…The following discrete time model was considered (here exemplified for the FitzHugh Nagumo model) [13]. A detailed treatment of the model and its discrete-time implementation is given in [13], Here we remind the defining equations of the RD-CNN cell: (2)…”

“…A detailed treatment of the model and its discrete-time implementation is given in [13], Here we remind the defining equations of the RD-CNN cell: (2)…”

A Python Framework for Fast Modelling and Simulation of Cellular Nonlinear Networks and other Finite-difference Time-domain Systems

“…The following discrete time model is considered in this paper (here exemplified for the FitzHugh Nagumo model) [14]. The equations defining the RD-CNN model are: The entire set of the RD-CNN model parameters is called a gene and will be allocated to a single variable Pars as shown next.…”

“…Other models can be adapted straightforwardly, for instance the Cellular Neural Network (CeNN) [13] which, in the context of emerging artificial intelligence (AI) applications can be regarded as a recurrent convolutional neural network (CNN). Following guidelines in [14] and [15] where a NUMBA implementation was first proposed, and using the proposed MSF as a template, one can easily adapt it for the CNN or other models of interest.…”

Automatic detection of skin melanoma from images using natural computing approaches

“…Although in [6] the continuous time model is presented, the one used in order to increase speed is the discrete time model, whose implementation is detailed in [7]. A C++ implementation was developed in [8] ensuring processing times in the order of seconds.…”