In this talk I will present a new type of on-line Competitive Learning (CL) networks, namely, Adaptive Competitive Learning Network (ACLN), which contains both feed-forward and feedback signals. The feed-forward signals have the same function as those in conventional CL networks, while the feedback processes transfer signals when classified patterns are removed. To construct ACLNs, we develop a new on-line adaptive CL method called Adaptive Self-spawning Competitive Learning (ASCL). Different from One-Prototype-Take-Multi-Cluster (OPTMC) learning strategy in off-line clustering algorithms, ASCL adopts the One-Prototype-Take-One-Cluster (OPTOC) learning strategy and on-line locates clusters one by one. In addition, ASCL is insensitive to initialization and robust to noise. We describe the evolution process of the architecture of ACLN and propose the ASCL-based construction algorithm. Our ACLN and ASCL are applied to pattern clustering. In clustering analysis, our results show that ASCL efficiently adapts to pattern variations and is able to accurately locate the non-degenerated clusters in simulated adaptive data sets.

Professor Liu, Zhi-Qiang has worked in both academic and industry. He has taught in three Universities in Canada, the University of Melbourne, Australia, and City University of Hong Kong, SAR, China.

Professor Liu was a Senior/Principal systems engineer in electronics and communications industry in China and Canada. He led the following major research and development projects: China's first generation of inertia-based guidance systems for navigation of missiles and aircraft; Automated manufacturing systems; Channel equalization schemes in wireless and mobile communications; CDMA coding scheme initiatives (written in C programming language), which was supported in part by the Canadian Government; Wireless data communications, in particular, for mobile office scenarios; and Internet/Intranet wireless data communications in enterprises.