Immune-inspired incremental feature selection technology to data streams

As data streams are gaining prominence in a growing number of emerging applications, advanced analysis and mining of data streams is becoming increasingly important. In this paper, an immune-inspired incremental feature selection algorithm called ISFaiNET is proposed as a solution for mining data streams, immune network memory antibody set which is far less than the size of data streams is design as a sketch data set. We can get the change features to the most extent by this set. ISFaiNET have the ability of feature extraction of dynamically tracking increasing huge size information by introducing increment strategy such as window mechanism. The empirical results for our algorithm are presented and discussed which demonstrate acceptable accuracy coupled with efficiency in running time.