Intelligent Hybrid Encryption Framework Using Machine Learning
Main Article Content
Abstract
Modern data is dispersed across cloud, IoT, mobile, and business contexts. Data breaches can still be caused by human error, even with A.I. in cybersecurity. The Intelligent Hybrid Encryption Framework (IHEF) uses machine learning to pick encryption algorithms depending on file attributes, security needs, and system resources. Five base classifiers (XGBoost, LightGBM, CatBoost, Random Forest, Extra Trees) are integrated via a stacking ensemble using logistic regression as the meta-learner to predict one of five algorithms: AES-128, AES-256, ChaCha20, RSA-2048, and ECC. A synthetic dataset of 10,000 records with contextual features (file size, sensitivity, threat level, network latency, CPU/memory usage, battery level, security score, throughput) yielded strong results: stacking ensemble achieved 99.1% accuracy, and the best model achieved ~98.8% per-class accuracy (macro-F1 98.5%, weighted-F1 98.8%). Confusion-matrix analysis shows few class misclassifications, and feature-importance / explain ability analyses show security-related and system-resource features drive algorithm selection. This suggests that IHEF can facilitate encryption selection in dynamic cybersecurity contexts using an adaptive and explainable mechanism that balances predictive performance with system resource considerations. computationally efficient mechanism.
Article Details

This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.