This paper presents an innovative approach to the development of a semi-supervised Support Vector Machine aimed at classifying radio frequency signals in the communication systems of unmanned aerial vehicles. It assumes the possibility of overlapping distributions of different types of signals. The cost function is modified by introducing penalty elements for misclassification based on a linear function of the distance between the signal and the classification hyperplane. A voting method using different kernels is employed to integrate the predictions from various models. The optimization of hyperparameters is carried out using the Optuna framework.
Semi-Supervised Support Vector Machine for Classification of Radio Frequency Signals in Unmanned Aerial Vehicles
2024-10-22
310636 byte
Conference paper
Electronic Resource
English
Detection and classification of unmanned aerial vehicles
European Patent Office | 2024
|DETECTION AND CLASSIFICATION OF UNMANNED AERIAL VEHICLES
European Patent Office | 2022
|