Outlier is the data object which does not comply with the general behaviour or model of data. Which are grossly different from entire set of data. From large data set detecting outliers present different challenge resulting from curse of dimensionality. As the data size is double every year, there is a need to detect outlier in large datasets as early as possible. If there are lot of outliers in data set there might de misclassification of data and outlier data might be classified as normal data. More contrasting outlier score gives by SVM in high dimensional data in which training the data set is relatively easy. SVM mainly focusing on high dimensionality of data, this method will be allowed to use a training data set to train the classifier while detecting outliers from high dimensional data.