.png){kind=icon}
Return
Time-frequency analysis using spiking neural network
.png){kind=icon}
OA
AI
DOI：10.1088/2634-4386/ad80bc.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Time-frequency analysis plays a crucial role in various fields, including signal processing and feature extraction. In this article, we propose an alternative and innovative method for time-frequency analysis using a biologically inspired spiking neural network (SNN), encompassing both a specific spike-continuous-time-neuron-based neural architecture and an adaptive learning rule. We aim to efficiently detect frequencies embedded in a given signal for the purpose of feature extraction. To achieve this, we suggest using an SN-based network functioning as a resonator for the detection of specific frequencies. We developed a modified supervised spike timing-dependent plasticity learning rule to effectively adjust the network parameters. Unlike traditional methods for time-frequency analysis, our approach obviates the need to segment the signal into several frames, resulting in a streamlined and more effective frequency analysis process. Simulation results demonstrate the efficiency of the proposed method, showcasing its ability to detect frequencies and generate a Spikegram akin to the fast Fourier transform (FFT) based spectrogram. The proposed approach is applied to analyzing EEG signals, demonstrating an accurate correlation to the equivalent FFT transform. Results show a success rate of 94.3% in classifying EEG signals.
Keywords:
SNN
STDP
neuromorphic computing
time-frequency analysis
feature extraction
frequency detection
EEG.
Journal
IF:
6.1
Papers:
340
Citations:
920
Organization
Cited Papers
ShareSaveShareSaveBarriers and facilitators of sports in children with physical disabilities: a mixed-method study
PREAI
ShareSaveShareSaveOverview of Spiking Neural Network Learning Approaches and Their Computational Complexities.png)
SENSORS
OAAI
SENSORSIF3.5
OAAIShareSave
View more