Network Log Anomaly Detection Based on GRU and SVDD
Shirong Liu, Xiong Chen, Xingxiong Peng, Ruliang Xiao · 2019
Using machine learning to detect anomalies in network logs has become a research hotspot in the field of industrial Internet of Things security. In the era of large data, it is inefficient using traditional methods to detect anomalies under the environment of high-dimensional data and extensive data association. How to detect anomalies efficiently and accurately is a challenge. This paper presents a novel method of anomaly detection for network logs based on GRU and SVDD. First, PCA is used to reduce the dimension of high-dimensional datasets and extract effective attributes. Then, the processed datasets are used to train the GRU-SVDD classifier model. Finally, the actual logs to be detected are input into the GRU-SVDD comparator to detect the day. Experiments on classical KDD Cup99 datasets show that our method is superior to classical GRU-MLP and LSTM algorithms.