The following information was submitted:
Transactions: WSEAS TRANSACTIONS ON INFORMATION SCIENCE AND APPLICATIONS
Transactions ID Number: 42-307
Full Name: Guisong Liu
Position: Associate Professor
Age: ON
Sex: Male
Address: School of Computer Science and Engineering University of Electronic Science and Technology of China
Country: CHINA
Tel:
Tel prefix:
Fax:
E-mail address: Guisongliu@gmail.com
Other E-mails: lgs@uestc.edu.cn
Title of the Paper: Modified Neural Gas Algorithm for One-Class Pattern Construction
Authors as they appear in the Paper:
Email addresses of all the authors:
Number of paper pages: 10
Abstract: Traditional Neural Gas (NG), as a soft competitive neural method, can be applied for clustering analysis effectively. However, it is not able to characterize one-class pattern due to its static structure and presentation of neurons. This paper proposes a Modified Neural Gas (MNG) algorithm, which aims to overcome the shortages of the traditional NG. By a predetermined splitting radius ¦Ë, the neuron number can be determined adaptively and a certain class pattern can be constructed by many sub-patterns represented by each neuron in the MNG network. The class boundaries will become more accurate by competitive and cooperative processes. Simulations are carried out to show the performance of the proposed algorithm by using synthetic datasets and the KDD CUP 1999 intrusion detection evaluation datasets.
Keywords: Neural gas network, Soft computing, Cluster analysis, Anomaly detection
EXTENSION of the file: .doc
Special (Invited) Session:
Organizer of the Session:
How Did you learn about congress:
IP ADDRESS: 125.69.143.40