EEG Signal Processing Based on Incremental Sparse Nonnegative Matrix
Qiuyue Li, Guo Shuli, Han Lina · 2023
This paper proposes a novel algorithm for the classification identification of epileptic electroencephalogram (EEG) signals. The method utilizes an incremental sparse nonnegative matrix and feature selection to improve the accuracy of classification between seizure and nonseizure signals. The proposed algorithm, based on a sparse incremental non-negative matrix (INMFSC), models the sparse characteristics of EEG signals and enhances online learning efficiency. Simulation experiments demonstrate that INMFSC achieves faster processing speed while improving classification recognition accuracy for epileptic EEG signals. Furthermore, INMFSC exhibits superior performance in distinguishing seizure and nonseizure phases of epileptic patients. This method provides a benchmark for further investigation development of algorithms for the analysis and classification of epileptic EEG signals.