Detection of ventricular fibrillation (VF) and rapid ventricular tachycardia (VT) is crucial for the success of the defibrillation therapy. A novel detection algorithm that combines ECG parameters with SVM to identify VF/shockable arrhythmias has been presented. SVM learning algorithms can improve the efficiency for the detection of life-threatening arrhythmias. FS techniques might help to provide valuable insights to build highly accurate detection algorithms. Prompt detection of VT and VF episodes is crucial to deliver an electric shock therapy and in this way increase the probability of survival from a SCA incident. The development of automated external defibrillators that analyze the surface electrocardiogram (ECG) signal if either rapid VT or VF is detected. Ventricular fibrillation (VF) versus ventricular Tachycardia (VT), making it difficult to assess the real performance. In this previous method have used genetic algorithms (GA) or discriminant analysis. The performance of the proposed SVM detection used and combined three different FS filter-type techniques into a single ranking score, allowing us to determine the relevance of each ECG parameter. Using this score in the SVM classifier to yield a robust result. In this approach Hurst index and the PSR parameter was calculated. Case of SVM classifiers using unbalanced datasets which provide arrhythmia detection problems. Temporal/Morphological Parameters, Spectral parameters and Complexity parameters have been measured with the filter technique. The BER metric is an interesting magnitude to set the free parameters of the algorithm. Detection performance of the analyzed parameters, this procedure is an interesting method for evaluating the discrimination ability of a set of ECG parameters performed well. This SVM algorithms combining ECG features significantly improves the efficiency for the detection of life-threatening arrhythmias. FS method has been combined to the SVM algorithm to provide a robust classifier using a reduced set of ECG parameters. In future we implement multicast SVM classifier with ad boost. A SVM is a binary classifier, that is, the class labels can only take two values:±1. Binary Tree Architecture (SVM-BTA) takes advantage of both the efficient computation of the tree architecture and the high classification accuracy of SVMs Value can also be interpreted as a confidence value. The larger the value the more confident one is that the point x belong to the positive class. Enhance the binary clarification type of normal and abnormal. In multiple classifications the types divide into the normal stage, moderate stage, beginning stage or severe stage.