Artificial Neutral Networks (ANN) based classification of the electrocardiogram signal to identify various diseases of the heart is a popular method. However, as the classification error can lead to fatal consequences, so the new approaches to improve the accuracy of ANN based methods has been an active area of research during the past decade. An ANN ensemble based approach to detect the arrhythmia from ECG signal by classifying into normal and abnormal classes has been proposed in this presented work. The high accuracy of classification has been observed from the results which makes the proposed approach very effective. Another advantage of the proposed approach is its straightforward design which makes it very easy for implementation.