Fault‐Tolerant Transformer‐Based Blockchain Authentication and Access Control With Snow Geese Optimization for Hybrid Neural Network‐Based Intrusion Detection

ABSTRACT

The field of network security has greatly advanced due to the implementation of intrusion detection systems (IDS), which are capable of identifying unforeseen threats within network traffic. Contemporary techniques for detecting anomalies in networks predominantly utilize traditional machine learning models. The proposed Enhanced Fault‐Tolerant Transformer‐Based Slimmable Pruned Neural Network with Snow Geese Optimization (EFT‐SPNet‐SGO) to filter network intrusion is integrated with OPBFTCM based authentication, statistical decimal scaling normalization (SDSN), scale‐aware modulation meet transformer (SMMT) with feature extraction and slimmable pruned neural network (SPNet) with classification using snow geese optimization (SGO). The OPBFTCM is adopted for the authentication process. The data is then preprocessed using SDSN. Then, relevant features are extracted using the SMMT. Finally, the detection and categorization of network intrusions is carried out by the SPNet, which is optimized using the SGO algorithm. The efficacy of the proposed strategy can be assessed using metrics such as recall, accuracy, precision, F1‐score, latency, and comparability to other approaches. The experiment's findings demonstrate that the designed model outperformed, with precision, f‐measure, accuracy, and recall values obtained from the network security laboratory‐knowledge discovery and data mining (dataset) all reaching 99.3%.