Gexiang Zhang

Radar Emitter Signal Recognition Based on Resemblance Coefficient Features

Resemblance coefficient (RC) feature extraction approach for radar emitter signals was proposed. Definition and properties of RC were given. Feature extraction algorithm based on RC was described in detail and the performances of RC features were also analyzed. Neural network classifiers were designed. Theoretical analysis results and simulation experiments of 9 typical radar emitter signal feature extraction and recognition show that RC features are not sensitive to noise and average accurate recognition rate rises to 99.33%, which indicates that the proposed approach is effective.

A hybrid classifier based on rough set theory and support vector machines

Rough set theory (RST) can mine useful information from a large number of data and generate decision rules without prior knowledge. Support vector machines (SVMs) have good classification performances and good capabilities of fault-tolerance and generalization. To inherit the merits of both RST and SVMs, a hybrid classifier called rough set support vector machines (RS-SVMs) is proposed to recognize radar emitter signals in this paper. RST is used as preprocessing step to improve the performances of SVMs. A large number of experimental results show that RS-SVMs achieve lower recognition error rates than SVMs and RS-SVMs have stronger capabilities of classification and generalization than SVMs, especially when the number of training samples is small. RS-SVMs are superior to SVMs greatly.

Resemblance Coefficient and a Quantum Genetic Algorithm for Feature Selection

Feature selection is always an important and difficult issue in pattern recognition, machine learning and data mining. In this paper, a novel approach called resemblance coefficient feature selection (RCFS) is proposed. Definition, properties of resemblance coefficient (RC) and the evaluation criterion of the optimal feature subset are given firstly. Feature selection algorithm using RC criterion and a quantum genetic algorithm is described in detail. RCFS can decide automatically the minimal dimension of good feature vector and can select the optimal feature subset reliably and effectively. Then the efficient classifiers are designed using neural network. Finally, to bring into comparison, 3 methods, including RCFS, sequential forward selection using distance criterion (SFSDC) and a new method of feature selection (NMFS) presented by Tiejun Lu are used respectively to select the optimal feature subset from original feature set (OFS) composed of 16 features of radar emitter signals. The feature subsets, obtained from RCFS, SFSDC and NMFS, and OFS are employed respectively to recognize 10 typical radar emitter signals in a wide range of signal-to-noise rate. Experiment results show that RCFS not only lowers the dimension of feature vector greatly and simplifies the classifier design, but also achieves higher accurate recognition rate than SFSDC, NMFS and OFS, respectively.

Quantum computing based machine learning method and its application in radar emitter signal recognition

Feature selection plays a central role in data analysis and is also a crucial step in machine learning, data mining and pattern recognition. Feature selection algorithm focuses mainly on the design of a criterion function and the selection of a search strategy. In this paper, a novel feature selection approach (NFSA) based on quantum genetic algorithm (QGA) and a good evaluation criterion is proposed to select the optimal feature subset from a large number of features extracted from radar emitter signals (RESs). The criterion function is given firstly. Then, detailed algorithm of QGA is described and its performances are analyzed. Finally, the best feature subset is selected from the original feature set (OFS) composed of 16 features of RESs. Experimental results show that the proposed approach reduces greatly the dimensions of OFS and heightens accurate recognition rate of RESs, which indicates that NFSA is feasible and effective.

Discretization of continuous attributes in rough set theory and its application

Existing discretization methods cannot process continuous interval-valued attributes in rough set theory. This paper extended the existing definition of discretization based on cut-splitting and gave the definition of generalized discretization using class-separability criterion function firstly. Then, a new approach was proposed to discretize continuous interval-valued attributes. The introduced approach emphasized on the class-separability in the process of discretization of continuous attributes, so the approach helped to simplify the classifier design and to enhance accurate recognition rate in pattern recognition and machine learning. In the simulation experiment, the decision table was composed of 8 features and 10 radar emitter signals, and the results obtained from discretization of continuous interval-valued attributes, reduction of attributes and automatic recognition of 10 radar emitter signals show that the reduced attribute set achieves higher accurate recognition rate than the original attribute set, which verifies that the introduced approach is valid and feasible.

Quantum evolutionary algorithm for multi-objective optimization problems

In this paper, a novel evolutionary algorithm called new quantum evolutionary algorithm (NQEA) is proposed to solve a class of multi-objective optimization problems. The main point of NQEA is that a new quantum logic rotation gate is introduced. NQEA characterizes rapid convergence, good global search capability and short computing time. Then, the convergence of NQEA is also analyzed using random functional theory. The results from optimization design of IIR digital filters demonstrate that NQEA is superior to other several conventional evolutionary algorithms greatly in quality and efficiency.

Support vector machines with huffman tree architecture for multiclass classification

This paper proposes a novel multiclass support vector machine with Huffman tree architecture to quicken decision-making speed in pattern recognition. Huffman tree is an optimal binary tree, so the introduced architecture can minimize the number of support vector machines for binary decisions. Performances of the introduced approach are compared with those of the existing 6 multiclass classification methods using U.S. Postal Service Database and an application example of radar emitter signal recognition. The 6 methods includes one-against-one, one-against-all, bottom-up binary tree, two types of binary trees and directed acyclic graph. Experimental results show that the proposed approach is superior to the 6 methods in recognition speed greatly instead of decreasing classification performance.

A novel feature selection approach and its application

Feature selection is a satisfactory optimization problem. Most feature selection methods did not consider the cost of feature extraction and the automatic decision of feature subset dimension. So a novel approach called satisfactory feature selection method (SFSM) was proposed. SFSM integrated feature extraction with feature selection and considered classification performance of feature samples, the dimension of feature subset and the complexity of feature extraction simultaneously. Experimental results show that SFSM selects more satisfying feature subset than sequential forward selection using distance criterion (SFSDC) and the method presented by Tiejun Lu (GADC). Also, SFSM achieves higher accurate recognition rate than original feature set, SFSDC and GADC, which verifies the validity of the proposed method.

A novel approach for radar emitter signal recognition

To enhance recognition rate of radar emitter signals (RESs) to meet the requirements of modem electronic warfare, a novel approach is proposed to recognize automatically different RESs. The main points of the introduced approach include resemblance coefficient feature extraction and support vector machine classifiers. Experimental results show that the proposed approach is a good method for recognizing RESs and support vector machines are better classifiers for solving pattem recognition problems than neural networks.

Radar emitter signal recognition based on feature selection algorithm

Rough set theory (RST) was introduced into radar emitter signal (RES) recognition A novel approach was proposed to discretize continuous interval valued features and attribute reduction method was used to select the best feature subset from original feature set Also, rough neural network (NN) classifier was designed Experimental results show that the proposed hybrid approach based on RST and NN achieves very high recognition rate and good efficiency It is proved to be a valid and practical approach.