A Hybrid Method for Anomaly Detection Using Distance Deviation and Firefly Algorithm
Aparna Shrivastava, Potukuchi Raghu Vamsi · 2023
This paper proposes an outlier detection algorithm that combines distance based outlier detection method with the optimization by swarm intelligence for detection of anomalies in the large sensor data. To detect outliers, the algorithm calculates the Multi-granularity DEviation Factor (MDEF) of each data point from all its neighbors which lie within a circle of radius r. The basic idea of the proposed method is to use a self-optimizing algorithm to find the best value of distance (radius) over which MDEF calculation is done for all data point to correctly identify outliers from data. The proposed technique does not only identify the outliers in the data but also gives a score to indicate the degree to which a point deviates from normal data. The proposed method has been simulated using well known real world data set. The results show that the proposed method is more accurate when compared to other outlier detection algorithm.