An Optimized LOF algorithm Based on Tree structure

LOF (Local Outlier Factor) is a classical local outlier detection algorithm, widely used in network intrusion detection and processing of classification benchmark data. However, the algorithm spends too much time on computing, searching and querying data objects, which leads to high time complexity and poor scalability of the algorithm, and cannot complete outlier detection for large-scale data sets. In order to solve the above problems, this paper proposes an improved LOF algorithm based on tree structure, which is called ILOF (Increment Local Outlier Factor), and uses a tree structure with a more efficient way of building and querying. The results of the experiment indicate that ILOF not only ensures the detection accuracy, but also improves the efficiency of outlier detection and has good scalability.