Edge-to-Cloud Synergy: An Autoencoder-GAN Framework for Anomaly Detection in Healthcare Records, Financial Statements, and Secure Cloud Storage

Information Technology

Anomaly detection is a key function in protecting and managing data in sensitive environments such as healthcare, finance, and cloud storage networks. This paper presents an Edge-to-Cloud synergy model that leverages the strengths of both edge and cloud computing to attain efficient and scalable anomaly detection. The architecture uses a shallow autoencoder at the edge level for initial anomaly filtering and bandwidth optimization to ensure data privacy and responsiveness. Only suspicious marked information or summary data is transmitted to the cloud and not everything, in order to save on transmission and reduce latency. Within the cloud layer, an advanced Autoencoder-GAN framework is applied to perform deep anomaly detection. The autoencoder captures normal behavioral patterns and tags anomalies with high reconstruction error, while the GAN improves detection by learning subtle data distributions and generating adversarial samples to further improve robustness. The combined framework ensures accurate detection, especially for high-dimensional, heterogeneous, and imbalanced data. The method begins with data collection from three key sources: anonymized EHRs for healthcare, SEC EDGAR filings for economic data, and AWS CloudTrail logs on Kaggle for cloud transactions. Pre-processing includes cleaning, normalization, encoding, and partitioning into training, validation, and test sets. Everything in the pipeline was implemented in Python using TensorFlow, Keras, and Scikit-learn libraries. Experimental testing confirms that the proposed model is superior to traditional methods, with high precision, recall, and accuracy. The last model obtained a 95.8% accuracy in anomaly detection, confirming the performance of the Edge-to-Cloud Autoencoder-GAN framework.

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