Remarks on estimating upper limit of norm of delayed connection weight matrix in the study of global robust stability of delayed neural networks

The question of estimating the upper limit of the norm ∥B∥2 of the delayed connection weight matrix B, which is a key step in some recently reported global robust stability criteria for delayed neural networks (DNNs), is considered. An estimate of the upper limit of ∥B∥2 was previously given by Cao, Huang and Qu. More recently Singh has presented an alternative estimate. Presently it is shown that an estimate of the upper limit of ∥B∥2 may be found in some cases, which would be an improvement over each of the above-mentioned two estimates. Some observations concerning the determination of the least conservative upper limit of ∥B∥2 are presented.