Contents I. Introduction
Over the past several decades, extensive applications in artificial intelligence and realistic engineering have enormously accelerated the interrelated research on neural networks in both continuous-time and discrete-time domains, and tremendous significant results have been reported, involving the stability analysis [1]–[5]; state estimation [6]–[8]; filter design [9], [10]; synchronization control [11]–[14]; and passivity and dissipativity synthesis [15]–[18]. Apart from this, switched neural-networks, as the important branch, have also received considerable attention in the neural network community. The primary reasons that yield the present circumstances stem from the following three factors: 1) the successful applications in high-speed signal processing and the gene selection in DNA microarray analysis [20]; 2) the extraordinary capability in processing the associative memory and optimization problems [19]; and 3) the switching characteristics caused by the unexpected variations in neural-network structures and information latching [21]. Meanwhile, with the arbitrary switching or the time-dependent switching mechanisms (containing the dwell-time, average dwell-time, mode-dependent average dwell-time, and persistent dwell-time switching paradigms), recent years have witnessed the comprehensive investigations on the switched neural networks [6], [9], [20]–[28] by means of the corresponding switched system theories [29]–[36].
