基于改进TSVM的通信辐射源个体识别

为了解决通信辐射源个体中标签获取难问题，引入半监督机器学习理论，提出了一种基于预测置信度进行迭代的半监督学习算法(Improved Transductive Support Vector Machine Iterative Algorithm Based on the Confidence of Prediction，CP-TSVM)。该方法在TSVM算法的基础上，充分利用无标签样本，根据预测结果置信度进行迭代，能够大幅度减少分类器的运算量。计算机仿真表明，在有标签样本数目占总样本2%的情况下，CP-TSVM较TSVM算法在保证识别准确率的同时，模型训练时间缩短近60 s。

Individual Identification of Communication Emitter Based on Improved Transductive SVM

To cope with the problem that individual identification of communication emitter suffers from the lack of labeled samples,the semi-supervised machine learning theory is introduced,and a semi-supervised learning iterative algorithm based on confidence of prediction(CP-TSVM) is proposed.The algorithm iterates according to the reliability of the predicted results on the basis of TSVM,which can greatly reduce the computation of the classifier.In the data set of radiation source signal generated by simulation,CP-TSVM can guarantee the recognition accuracy and shorten the model training time nearly 60 s,compared with the TSVM algorithm when the number of labeled samples accounted for 2% of the total samples.