CMOS realization of a 2-layer CNN universal machine chip

Discrete-Time Cellular Neural Networks in FPGA

“…This greatly facilitates the implementation of CNN with hundreds or thousands of millions of neurons in a multi-FPGA system. These characteristics confer the proposed architecture a great advantage against other implementations in ASIC (e.g [2]) and FPGA (e.g. [3]), where the number of cells is considerably smaller.…”

A cellular neural network based method for classification of magnetic resonance images: Towards an automated detection of hippocampal sclerosis

“…Interactions between the cells of a CNN are only local and usually translation invariant, i.e., a connection from a cell j towards another cell i only exists if j is part of i's neighborhood and its type and strength depend only on the relative position of j with respect to i. Thus, the number of connections increases only linearly with the number of cells, a property that already enabled hardware realization of CNN (Cruz and Chua, 1992;Espejo et al, 1996;Carmona et al, 2003;Rodriguez-Vazquez et al, 2005;Flak et al, 2006) (see (Roska and Rodriguez-Vazquez, 2002;Roska, 2005Roska, , 2007 for an overview), as opposed to other types of ANN.…”

CNN‐based trajectory analysis of flagellar bacteria for nanoscale motion control

“…The ÿnal aim is to apply non-linear techniques in order to establish if the movements of the microorganism are characterized by deterministic e ects (chemotaxis) or by Brownian noisy e ects. The experimental setup is based on cellular nonlinear networks (CNNs) [5] and its VLSI implementation (ACE4k [6], ACE16k [7] and CACE1k [8] chips).…”