Loading [MathJax]/extensions/MathMenu.js
Research Progress and Prospects of Pre-Training Technology for Electromagnetic Signal Analysis | IEEE Journals & Magazine | IEEE Xplore

Research Progress and Prospects of Pre-Training Technology for Electromagnetic Signal Analysis


Contrastive pre-training learning is a method that combines pre-training learning and contrastive learning, aimed at distinguishing similar and dissimilar instances. The ...

Abstract:

The pre-training technology can overcome the problem of label dependence and improve the generalization ability of the model, it has achieved remarkable results in natura...Show More

Abstract:

The pre-training technology can overcome the problem of label dependence and improve the generalization ability of the model, it has achieved remarkable results in natural language processing (NLP) and computer vision (CV), and extensive exploration has been carried out in the analysis of time series data (TS) and spatio-temporal data (STD). Electromagnetic signal is a special time series, and the time series analysis pre-training technology has important reference significance for the electromagnetic signal analysis pre-training. Therefore, this survey compares and analyzes the time series data and electromagnetic signals, summarizes time series analysis pre-training methods, and reviews the research progress of electromagnetic signal analysis pre-training technology. The open source data of electromagnetic signals are comprehensively summarized, which provides a massive data basis for the construction of the pre-training basic model for electromagnetic signal analysis, and the performances of the pre-training technology for electromagnetic signal analysis are evaluated. In addition, the challenges and future development directions of electromagnetic signal analysis pre-training technology are analyzed.
Contrastive pre-training learning is a method that combines pre-training learning and contrastive learning, aimed at distinguishing similar and dissimilar instances. The ...
Published in: IEEE Access ( Volume: 13)
Page(s): 54130 - 54147
Date of Publication: 21 March 2025
Electronic ISSN: 2169-3536

Funding Agency:


References

References is not available for this document.