Fast Signal Clustering Based on a Compound Adaptive Resonance Model

摘要：

Design a new algorithm for fast signal clustering based on ANN via compound adaptive resonance model. The compound model use modular architecture, and its input layer is composed of several subordinated artificial neural networks, which extract each feature of data sets to form knowledge remember after supervised learning, and convert the original input data into inner normal input on learned knowledge base, to reduce the complexity and calculated amount of adaptive resonance when running. The pattern matching and unsupervised learning in adaptive resonance procedure are all based on comprehensible logic rulers. The system has obvious extensibility and evolutionary. Comparative simulation experiments are used to investigate the performance of new algorithm and ART2 on signal clustering problem. The result indicates that the new algorithm generally exhibits faster learning and high tolerance to noise interference. And it decreases the probability of false conviction when low vigilance values are used in case of low SNR.

关键词： artificial neural network(ANN) compound adaptive resonance model(CARM) signal clustering ELINT

作者: Liao Huirong Li Guolin Mao Weiping Zhou Wensong

作者单位: Graduate Students Brigade, Naval Aeronautic and Astronautic University, NASU Yantai, China Training Department, Naval Aeronautic and Astronautic University, NASU Yantai, China

会议类型: 国际会议

会议名称: 2010 International Conference on Information Security and Artificial Intelligence(2010年信息安全与人工智能国际会议 ISAI 2010)

会议地点: 成都

会议语种:英文

页码: 675-679

在线出版日期: 2010-12-17（万方平台首次上网日期，不代表论文的发表时间）