Auto-encoder anomaly detection method based on comparative learning

Abstract

The invention discloses an autoencoder anomaly detection method based on comparative learning. The method comprises the following steps: firstly, carrying out encoding feature extraction on an input normal sample; constructing and updating a feature storage module; abnormal disturbance is added through the multi-scale noise and the texture data set, and an abnormal sample is generated; performing multiple groups of enhancement operations on the abnormal sample data, combining the abnormal sample data with normal samples, and making negative sample pairs required by a contrast learning framework; reconstructing an abnormal sample through an auto-encoder, and calculating an error before and after image reconstruction according to comparison loss; in the detection stage, reconstruction similar to training data is obtained; and determining whether the input data is abnormal or not through the evaluation system, and positioning to obtain a final anomaly detection result. According to the method, the characteristics of comparative learning are utilized, a reasonable positive and negative sample pair is constructed through the anomaly embedding module and the auto-encoder, meanwhile, the feature storage module enables normal samples to be better reconstructed and abnormal data reconstruction to be inhibited in the detection process, and the anomaly detection effect is effectively improved.

Description

technical field [0001] The invention belongs to the technical field of artificial intelligence, and in particular relates to a method for detecting abnormality of an autoencoder based on contrastive learning. Background technique [0002] In recent years, with the improvement of the automation degree of the manufacturing industry, the requirements for the automation of material quality inspection and less manual intervention are getting higher and higher. In order to evaluate the structural safety of industrial products, all anomalies need to be accurately detected to determine whether the product is qualified. Different data types have different defects, and the need to identify these multi-category and multi-target defects becomes a challenging task. Therefore, in order to meet industry standards and strictly control the quality range, it is necessary to conduct product inspections in advance to find and remove substandard products. [0003] Compared with normal data, th...

AI Technical Summary

Problems solved by technology

However, the current detection methods still cannot accurately describe normal and abnormal samples, so it is necessary to improve on the basis of reconstruction methods to increase the gap between normal data and abnormal data after reconstruction, so as to effectively detect anomalies

Images