Neural Network-Based Anomaly Detection for Securing Cloud Data Transactions

Abstract

The sheer revolution that cloud computing has brought in data storage and processing has also brought with it enormous security challenges particularly in maintaining data transactions’ security. This study provides an anomaly detection framework based on a neural network specifically to protect the cloud data transactions. The suggested architecture of the model includes a mix of hybrid architecture, a combination of Convolutional Neural Networks (CNN) and Bidirectional Long Short-Term Memory (Bi-LSTM) networks and an attention mechanism to capture the most relevant data as found in complex data streams. The CNN layer detects the patterns in space, while Bi-LSTM processes temporal dependencies from both directions, allowing thorough anomaly detection. The attention layer also enhances the precision of detection because the most important segments of data are prioritized dynamically. The model is deployed using TensorFlow and Keras API in python, having high accuracy and low false positive rate and hence has great potential when deployed in real-time. This method offers a smart, scalable solution to the security and integrity requirements of data transactions on the cloud against ever-changing cyber threats.