Comparison of subspace-based methods with AR parametric methods in epileptic seizure detection

Epilepsy seizure detection using eigen-system spectral estimation and Multiple Layer Perceptron neural network

“…FeaturesðbiÞ ¼ ½MeanðbiÞ; StdðbiÞ; MaxðbiÞ; Entro pyðbiÞ; MobilityðbiÞ (16) where bi is each of the selected sub-bands. In this equation, we used Log-Entropy to compute the Entropy(bi) parameter which is calculated using the following relation where s i is ith sample of sub-band.…”

“…Subasi et al compared subspace-based method with AR spectral estimation for classification of EEG signal into the two states of seizure and non-seizure [16]. In their work, AR methods including Burg, Yule-Walker, and pseudo-spectrum algorithms such as MUSIC were compared with each other in terms of their frequency resolution and the effects on the classification of the epileptic seizure state.…”

“…In this article, having the classification for three states of itcal, inter-ictal and normal states in mind, instead of employing the pseudo-spectrum directly to a classifier, as it was used in [16,17], the resulted pseudo-spectrum is divided into several sub-bands. Then, certain features are extracted from each sub-band.…”

Comparison of AR parametric methods with subspace-based methods for EMG signal classification using stand-alone and merged neural network models

“…Even if the signalto-noise ratio (SNR) is low, they can still provide good PSD estimations. The mentioned techniques are most applicable to signals that might be assumed as the composition of some specific sinusoids buried in noise [20][21][22][23][24][25][26][27][28][29].…”

“…Multiple signal classification (MUSIC) [20][21][22][23][24][25] and EIG [25][26][27][28][29] techniques are subspace−based techniques that are utilized for acquiring power spectral density (PSD) signal estimations.…”