Training Support Vector Data Descriptors Using Converging Linear Particle Swarm Optimization

摘要：

It is known that Support Vector Domain Description (SVDD) has been introduced to detect novel data or outliers. The key problem of training a SVDD is how to solve constrained quadratic programming (QP) problem. The Linear Particle Swarm Optimization (LPSO) is developed to optimize linear constrained functions, which is intuitive and simple to implement. However, premature convergence is possible with the LPSO. The LPSO is extended to the Converging Liner PSO (CLPSO), which is guaranteed to always find at least a local optimum. A new method using CLPSO to train SVDD is proposed. Experimental results demonstrate that the proposed method is feasible and effective for SVDD training, and its performance is better than traditional method.

关键词： Support Vector Domain Description Constrained Quadratic Programming Linear Particle Swarm Optimization Premature Convergence Converging Linear Particle Swarm Optimization

作者: Hongbo Wang Guangzhou Zhao Nan Li

作者单位: College of Electrical Engineering, University of Zhejiang,Hangzhou, Zhejiang Province, China

会议类型: 国际会议

会议名称: International Conference on Life System Modeling and Simulation,and International Conference on Intelligent Computing for Sustainable Energy and Environment(2010生命系统建模与仿真国际会议暨m2010可持续能源与环境智能计算国际会议)

会议地点: 无锡

会议语种:英文

页码: 196-204

在线出版日期: 2010-09-17（万方平台首次上网日期，不代表论文的发表时间）