A cellular neural network for peak finding in high-energy physics

Describes the hardware implementation of a 2D cellular neural network (CNN) performing an online clustering algorithm. After a general introduction to CNNs, we consider a 2D CNN that performs a cluster peak-finding algorithm in a matrix of cells mapping a sub-region of a calorimeter, a detector largely used in high-energy physics. The peaks of the energy clusters are found in one collision time of the particle bunches (96 ns). Some quantitative parameters are given to optimize the architecture of the CNN, which was implemented in a commercial field programmable gate array (FPGA).