The early detection of arrhythmia is very important

for the cardiac patients. This done by analyzing the

electrocardiogram (ECG) signals and extracting some features

from them. These features can be used in the classification of

different types of arrhythmias. In this paper, we present three

different algorithms of features extraction: Fourier transform

(FFT), Autoregressive modeling (AR), and Principal Component

Analysis (PCA). The used classifier will be Artificial Neural

Networks (ANN). We observed that the system that depends on

the PCA features give the highest accuracy. The proposed

techniques deal with the whole 3 second intervals of the training

and testing data. We reached the accuracy of 92.7083

compared to 84.4 for the reference that work on a similar

data.-The early detection of arrhythmia is very important

for the cardiac patients. This is done by analyzing the

electrocardiogram (ECG) signals and extracting some features

from them. These features can be used in the classification of

different types of arrhythmias. In this paper, we present three

different algorithms of features extraction: Fourier transform

(FFT), Autoregressive modeling (AR), and Principal Component

Analysis (PCA). The used classifier will be Artificial Neural

Networks (ANN). We observed that the system that depends on

the PCA features give the highest accuracy. The proposed

techniques deal with the whole 3 second intervals of the training

and testing data. We reached the accuracy of 92.7083

compared to 84.4　for the reference that work on a similar

data.