Efficient Classification Technique for Outlier Detection

Outliers are the data objects that clearly differ in their behavior from the normal data. Outlier detection mainly aims at finding these data objects. Outlier detection has become the major area of research in data mining. This plays a crucial role in data mining. Most of the methods used for outlier detection, consider the positive data and their behavior, and then the data violating the behavior are termed as outliers. Most of the time the data may be corrupted making it difficult to identify the data clearly. To handle this problem the paper provides an approach to improve the classification efficiency by generating the likelihood value for each data in the dataset. The kernel K-means clustering is used to compute the likelihood value which defines the membership value towards each class. The data with the value is subjected to classifier thus improving the accuracy in outlier detection.