Downloads: 140 | Views: 296

Research Paper | Computer Science & Engineering | India | Volume 5 Issue 7, July 2016 | Popularity: 6.6 / 10

Rate This Article!      

Cluster in High Dimensional Data to Detect Outlier

Sonali. A. Patil, Snehal. S. Thokale

Abstract: In computer world data should be secured. To find deviated data or fraud in data some techniques are introduced and that method is nothing but Outlier Detection. , it helps to improve find out frauds and intruders. Outlier detection techniques are used for batch system also but for huge data we have to more careful to find such type of data. So, to overcome this over sampling principle component analysis is used. By using Principal Component Analysis (PCA), it is helpful to find Outlier. In this system our aim to detect the presence of Outlier from a large amount of data using an online updating method. As we are using Oversampling Principal Component analysis (osPCA), there is no need to store data matrix or covariance matrix each time and thus approach is to find anomalies in online data stream or large amount of data problems. In our proposed system we capture only UDP packets will include. Along with this we are proposing algorithm for clustering.

Keywords: Detection of Outlier, online updating, oversampling, principal components analysis, TCP And UDP packets, DBSCAN cluster algorithm

Edition: Volume 5 Issue 7, July 2016

Pages: 144 - 147

Make Sure to Disable the Pop-Up Blocker of Web Browser

Click to Cite this Article

Similar Articles