Logo image
Back
Enhancing Cloud Infrastructure Security with GraphSAGE-Based Intrusion Detection Systems
Conference proceeding

Enhancing Cloud Infrastructure Security with GraphSAGE-Based Intrusion Detection Systems

G. Nanda Kishor Kumar, Shakir Syed, Amit Karbhari Mogal, Tahera Abid, Anvesh Perada and Sampathirao Suneetha
2025 International Conference on Intelligent Systems and Computational Networks (ICISCN), pp 1-6
24 Jan 2025
DOI: https://doi.org/10.1109/ICISCN64258.2025.10934650

Abstract

Accuracy Computational modeling Feature extraction genetic algorithm (GA) graph convolutional network (GCN) graphSAGE grasshopper optimization algorithm (GOA) Organizations Overfitting Smoothing methods Training Cloud Computing Genetic Algorithms Optimization
Cloud computing provides scalable and adaptable infrastructure however is significantly susceptible to cyber-attacks. This work presents a hybrid methodology that integrates the GOA and GA for feature selection, alongside the Residual GraphSAGE model for intrusion detection. The hybrid GOA-GA method enhances feature selection by optimizing the balance between exploration and exploitation, mitigating overfitting, and augmenting classification performance. The Residual GraphSAGE model alleviates over smoothing while effectively capturing node and edge interactions for precise intrusion classification. Assessment on a benchmark dataset reveals exceptional performance, attaining an accuracy of 91.74%, surpassing established models such as GCN and ResNet. This study emphasizes the need of hybrid optimization and sophisticated graph-based learning in improving cloud infrastructure security.

Metrics

1 Record Views

Details

Logo image