If clustering can be performed in a continuous system, there is potential usefulness in processing speed for larger scale clustering. We have proposed clustering method of coupled chaotic circuit networks with learning. In this study, we propose a method to extract the center of a cluster by stabilizing synchronization through learning in networks where synchronization is not stable due to chaos.