Investigations into the deep learning-based mimic choice approach

Abstract

The issue of false positives that are wrongly labeled as network attack behavior due to uneven mimic decision results commonly arises as a result of software and hardware differentiation.As a result, a deep learning-based mimic decision approach was put out.The deep semantic properties of a variety of normal response data were investigated from various executions by building an unsupervised autoencoder-decoder deep learning model, and its statistical rules were examined and condensed.Furthermore, the feedback optimization mechanism and the offline learning-online decision-making mechanism were created to address the false positive issue, which improves target system security resilience and allows for the accurate detection of network assaults.