A Novel Filter Based Ensemble Based Anomaly Detection Model for Uncertain Data

Due to the rapid growth of high speed network, the risk of credit-card attacks on the complex networks are also increases accordingly. Anomaly discovery from the database is a process of filtering uncertain features, so that it can be used wide variety of applications. Since the online distributed data is the communication between the remote client and the centralized server, it is difficult to predict the occurrence of an anomaly in the large distributed data. Anomaly detection on the complex data must take a long time due to the large number of features. A large number of anomaly prediction models have been implemented in the literature to find the anomaly patterns or features using association and classification techniques. Unfortunately some anomaly detection models over data mining cannot cover all the normal or abnormal features. Traditional approaches mainly focus on detecting relevant patterns from the trained data in order to estimate the test data instances. In this proposed work, the robust partition based classifier is implemented to find the topmost anomalies using attribute relationships. This model efficiently detects the anomaly features along with uncertain features with high true positive rate. Experimental results show that proposed approach has high computational efficiency and anomaly detection rate compared to traditional anomaly detection techniques.