Advanced Anomaly Detection in Cloud Security Using Gini Impurity and ML

Cloud security research has shown significant advances in terms of anomaly detection, with attention on cloud security. For instance, autoencoders were used to improve anomaly detection in cloud networks indicating that these methods can outperform traditional ones. In comparison with previous techniques, the proposed technique differs from them in the way it calculates reconstruction error; determines the appropriate thresholds and compares these errors against the thresholding method. In this research, grid search is used to tune model parameters and performance evaluations were then carried out using metrics such as accuracy, precision, recall, false positive rate and F1 score on CIDDS-001 dataset. Findings show that this could be a highly efficient approach for developing autoencoder-based IDS systems.