INTERNATIONAL JOURNAL OF CIRCUIT THEORY AND APPLICATIONS, VOL. 24,239 (1996) GUEST EDITORIAL In this second part of the Special Issue on Cellular Neural Networks the papers are arranged in the same areas as in the first one, which appeared in February of this year. As to theoretical aspects, qualitatively suprising effects are shown when using chaotic cells (Chua circuits) and oscillators as well as very simple template interactions. Generating n-double scrolls with CNNs shows how simple dynamic elements can be used to generate controllable complex spatiotemporal dynamics. A new way of creating spatial logic algorithms is introduced and a few prototype problems are solved by this technique including detecting the presence of bank-notes on a copier. In many applications the problem can be formalized in an abstract way to separate subspaces. This capability of CNNs is analysed in another paper. Some additional stability conditions are also presented, shrinking the set of useful stable templates whose stability we were unable to prove before. As to biologically inspired architectures, two papers are presented. It is shown partly as a tutorial to engineers, how fundamental interconnections and nerve activities can be modelled using CNNs. In addition, several spiking patterns of neurons are modelled by simple CNN cells and templates. Interesting VLSI implementation aspects are reported as well: an architecture of a digital VLSI implementation for vision applications, a circuit design technique for digitally programmed template values in analogue CNN arrays and an analysis of continuous-time VLSI-oriented CNN cell design methodology. Several successful applications are described. The robustness of texture classification and image deconvolution templates is analysed. It is shown that local inaccuracies in the template elements are not propagated; on the contrary, they are damped. 3D interpolation and approximation techniques and 2D object rotation templates are presented. Facial expression animation techniques are also introduced along with some visual illusion simulations. We hope that this double special issue will give readers ample opportunity to discover this new spatiotemporal computing paradigm where circuit techniques and computational capabilities are inherently coupled. The conventions and notations are the same as in the first special issue. Shortly after this issue has appeared, the fourth biennial Workshop on Cellular Neural Networks and Their Applications (CNNA-96) will start in Seville(23 June, 1996). We hope to find that many readers are stimulated by this special issue to participate actively in this workshop. Joos VANDEWALLE TAMASROSKA 0 1996 by John Wiley & Sons, Ltd.