Ioan CLEJU, Adriana SÎRBU, Cristian BONCIU
Hopfield Neural Network in Error-Correcting Codes Synthesis

Abstract.
In this paper we present a software package designed to synthesize error-correcting codes based on a Hopfield Neural Network approach. The codes synthesis problem is stated in terms of finding maximal cliques in the graph G(V,E), where V is the set vertices corresponding to all possible codewords and E is the set of edges corresponding to Hamming distances between two codewords greater than a specified threshold. This maximal clique problem is solved using a Hopfield-type neural network, with specific state-update rules.

Keywords: error-correcting codes, graph-theoretic codes, Hamming distance, Max-Clique algorithms, Hopfield neural network.