An Adaptive Self-Organizing Algorithm with Virtual Connection
Shingo Kawahara and Toshimichi Saito
EEE Dept., HOSEI Univ. 3-7-2 Kajino-cho, Koganei-shi, Tokyo 184-8584, Japan
In this paper, we propose an unsupervised learning algorithm that uses the concept of virtual connection. Performing some numerical experiments, we have confirmed that the algorithm has important functions: classification of the input space, labeling of the classified input space, and flexible adaptation to time-variant input space. The algorithm is suited for hardware implementation. An application to vector quantization is also discussed.