METHOD AND ALGORITHM FOR EXTRACTING FEATURES FROM DIGITAL SIGNALS BASED ON NEURAL NETWORKS TRANSFORMER

Abstract

Recently, neural network models have become one of the most promising directions in the field of automatic
feature extraction from digital signals. Traditional approaches, such as statistical, time-domain,
frequency-domain, and time-frequency analysis, require significant expert knowledge and often prove insufficiently
effective when dealing with non-stationary and complex signals, such as biomedical signals (ECG,
EEG, EMG) or industrial signals (e.g., currentgrams). These methods have several limitations when it comes
to analyzing multichannel data with varying frequency structures or when signal labeling is too laborintensive
or expensive. Modern neural network architectures, such as transformers, have demonstrated high
efficiency in automatic feature extraction from complex data. Transformers have outperformed traditional
convolutional and recurrent neural networks in many key metrics, particularly in tasks involving time series
forecasting, multimodal data classification, and feature extraction from sequences. Their ability to model
complex temporal dependencies and nonlinear relationships in data makes them ideal for tasks such as noise
filtering and multimodal signal processing. This paper proposes a method for feature extraction from digital
signals based on a modified transformer architecture that incorporates a nonlinear layer after the selfinspection
module. This approach improved the ability of the model to detect complex and nonlinear dependencies
in the data, which is particularly important when dealing with biomedical and signals obtained from
industrial systems. A description of the architecture and the experiments performed are presented, demonstrating
the high performance of the model in solving signal classification, prediction and filtering problems.
It is expected that the model can be applied to a wide range of applications including disease and fault
diagnosis, signal parameter prediction and system modelling.