Outliers detection based on negative selection algorithm

This paper presents a new approach to detect outliers. This paper detailedly introduces how to apply negative selection algorithm in outliers detection. Firstly, the maximum distance among all points is divided into a certain number of ranges which are encoded to binary codes. And then the distances between each point and a certain number (for example 20) points nearby are encoded to binary string based on the binary codes which were presented. Negative selection algorithm based on binary strings is applied to detect outliers. Experiments on random data to evaluate the effectiveness of the approach are presented. Experiments show that this approach can detect outliers effectively.