Unified Negative Selection Algorithm for Anomaly Detection

A novel negative selection algorithm is presented, which is inspired by the negative selection mechanism of the immune system that can detect foreign patterns in the complement (nonself) space. In the algorithm, the pattern space is unified into a certain interval and the foreign pattern detectors (in the complement space) are defined in the form of short intervals. Algorithm analysis reveals the bound of probability that detectors fail to detect an abnormal pattern and the bound of interval radius chosen to create a pattern interval. Experimental results show that the algorithm can generate detectors quickly and detect abnormal patterns effectively. These results also demonstrate the influence on algorithm performance when different pattern interval radiuses are chosen.