Optimal selection of time lags for TDSEP based on genetic algorithm

In this letter, a two-step learning scheme for the optimal selection of time lags is proposed for a typical temporal blind source separation (TBSS), Temporal Decorrelation source SEParation algorithm (abbreviated as TDSEP). Given the time lags, the time-delayed second-order correlation matrices are first diagonalized simultaneously. Then, a genetic algorithm is used to update the time lags. Finally, experimental results demonstrate that the proposed method can efficiently accomplish the aforementioned task.