Least square support vector data description for HRRP-based radar target recognition

• Published in: Applied intelligence (Dordrecht, Netherlands) Vol. 46; no. 2; pp. 365 - 372
• Main Authors: Guo, Yu, Xiao, Huaitie, Fu, Qiang
• Format: Journal Article
• Language: English
• Published: New York Springer US 01.03.2017; Springer Nature B.V
• Subjects: Aircraft; Artificial Intelligence; Classification; Computer Science; Constraint modelling; Datasets; Experiments; Intelligence; Least squares method; Machine learning; Machines; Manufacturing; Mathematical models; Mechanical Engineering; Processes; Radar targets; Recognition; Support vector machines; Target recognition; Least square; Support vector data description; High-resolution range profile; Machine learning; Radar automatic target recognition
• ISSN: 0924-669X; 1573-7497
• DOI: 10.1007/s10489-016-0836-5

A novel machine learning method named least square support vector data description (LSSVDD) is developed to classify the FFT-magnitude feature of complex high-resolution range profile (HRRP), motivated by the problem of radar automatic target recognition (RATR). The LSSVDD method not only inherits the advantage of LSSVM model, which owns low computational complexity with linear equality constraints, but also overcomes the shortcoming of poor capacity of variable targets in SVDD. Similar to the LSSVM, the distribution information within classes is found by least square method and applied for adjusting the boundary in LSSVDD, which relieves the over-fitting of SVDD. Hence, there will be a remarkable improvement in classification accuracy and generalization performance. Numerical experiments based on several publicly UCI datasets and HRRPs of four aircrafts are taken to compare the proposed method with other available approaches, and the results especially for multiple targets can demonstrate the feasibility and superiority of the proposed method. The LSSVDD is ideal for HRRP-Based radar target recognition.