Multi-index anomaly detection method based on neural network

Abstract

The invention discloses a multi-index anomaly detection method based on a neural network. The multi-index anomaly detection method comprises the following specific steps: 1, defining a data format; 2,carrying out model training on the system by utilizing SOM, and defining the system as a learning process; 3, performing anomaly detection on the input data, and defining the anomaly detection as a mapping process; and 4, when the model is mapped to be abnormal, carrying out root cause positioning. According to the multi-index anomaly detection method, the induction behavior model can be used forpredicting the unknown performance abnormality and providing an abnormality reason prompt, and the model can obtain higher prediction precision in a benchmark test result; the high-dimensional inputspace is mapped into the low-dimensional map space by using the SOM; and meanwhile, the topological property of the original input space is reserved, so that expandability and effective system behavior learning can be realized.

Description

technical field [0001] The invention relates to a technology in the computer field, in particular to a neural network-based multi-indicator anomaly detection method. Background technique [0002] An outlier is a data point that deviates sufficiently from other points to warrant suspicion of another mechanism. Anomaly detection methods have been used in various application domains, such as intrusion detection, financial fraud, medical diagnosis, law enforcement, and natural sciences. The most common outlier detection methods involve the use of distance-based methods, and despite their age, these methods have become the most popular and provide robust results. [0003] A particularly difficult case of anomaly detection is high-dimensional anomaly detection, where outliers are hidden by irrelevant attributes. In high-dimensional anomaly detection, many different methods such as feature bagging, high-contrast methods, statistical subspace selection, and spectral methods are us...

AI Technical Summary

Problems solved by technology

The learning scheme first needs to achieve scalability, which leads to a large learning overhead

Furthermore, system metrics for real distributed applications often fluctuate due to dynamic workload or measurement noise, which requires a robust learning scheme

Images