Research on Optimization Model of Biosignal Processing Feature Recognition System Based on Self-Supervised Learning

This study aims to build a bio signal processing feature recognition system based on self-supervised learning to improve the accuracy of signal processing and system optimization efficiency. First, through the algorithm design of the self-supervised learning model, the problems of large noise and uneven feature distribution in biological signals are solved, and the recognition accuracy of the model for features is enhanced. The model uses unlabeled data to improve the robustness of feature extraction by generating pseudo labels and multi-task learning. Secondly, this system uses a multi-layer convolutional neural network combined with an adaptive pooling layer to optimize the feature mapping process and reduce the interference of redundant data. In the system simulation link, the experiment uses a real biological signal data set and focuses on evaluating the accuracy, recall rate and computational efficiency of the model under different optimization strategies. The simulation results show that the system improves the feature recognition accuracy by 13.4% compared with the traditional supervised learning method in various indicators, and the processing speed is increased by 23.7%, achieving a significant optimization of computational efficiency based on ensuring accuracy.