An open set identification method and device for electromagnetic signals

Abstract

The invention discloses an open set identification method and device for electromagnetic signals, and the method comprises the steps: employing a sample set with a known signal category to train a convolutional neural network, obtaining a trained convolutional neural network, and obtaining a Weber distribution model according to a first coefficient output by a full connection layer and a signal category number; Inputting the to-be-identified electromagnetic signal into the convolutional neural network to obtain a second coefficient corresponding to the to-be-identified electromagnetic signal,and obtaining a third coefficient of an unknown signal category according to the Weber distribution model and the second coefficient; Obtaining an identification result of the identified electromagnetic signal according to the second coefficient, the third coefficient and the constructed open set classifier; According to the embodiment of the invention, the openmax open set classifier is introduced to fully utilize the distribution condition of the correctly detected sample parameters to construct the coefficient of the unknown sample to obtain the rejection model, and a better unknown samplerejection result is obtained under the condition that the recognition accuracy of the known sample is less reduced, so that the recognition reliability is ensured.

Description

technical field [0001] The invention relates to the technical field of signal identification, in particular to an open-set identification method and device for electromagnetic signals. Background technique [0002] With the rapid development of science and technology, electromagnetic signal recognition has a wide range of applications and important research value in national defense security, smart transportation, and security industries. Electromagnetic signal recognition in the general sense refers to giving the recognition result of the input signal by calculating the similarity between the input electromagnetic signal sample and the samples in the known signal library. Therefore, there are two types of recognition problems: 1) Close Set electromagnetic signal recognition without rejection, that is, it is assumed that the input electromagnetic signal sample must belong to an individual in the signal library; 2) Open set with rejection ( Open Set) identification, that is,...

AI Technical Summary

Problems solved by technology

Therefore, there are two types of recognition problems: 1) Close Set electromagnetic signal recognition without rejection, that is, it is assumed that the input electromagnetic signal sample must belong to an individual in the signal library; 2) Open set with rejection ( Open Set) recognition, that is, to judge whether the input electromagnetic signal sample belongs to the known signal library, and then give the recognition result on the basis of the determination

[0004] Although the above open set recognition methods reduce the false acceptance rate under the same false rejection rate, these methods only use the information of the dimension of the smallest distance or the largest confidence degree for open set discrimination, and discard the test samples and The distance or confidence between samples of each known class contains a large amount of discriminant information in the distribution in space

At the same time, the performance of the above open set recognition method is not satisfactory in the case of small samples.

Images