Analysis of Student Data Privacy Protection Based on Convolutional Neural Networks

ABSTRACT With the widespread adoption of digital platforms in education, large volumes of student data, ranging from personal details to behavioral logs, are continuously generated. While this data enhances learning personalization and administrative efficiency, it also raises significant concerns regarding privacy and data security. This research proposes a scalable walrus optimizer‐driven dynamic convolutional neural network (SWO‐DCNN) for analyzing and enhancing student data privacy protection. Data were collected that contain a synthetic student activity logs dataset comprising user interactions, access logs, and metadata. Two critical preprocessing steps were applied: Min–Max normalization for uniform feature scaling and one‐hot encoding to transform categorical variables into numerical formats suitable for deep learning (DL) models. The SWO algorithm enhances the DCNN's convergence speed and accuracy by adaptively tuning hyperparameters during training. The model leverages the optimization strength of the walrus algorithm to improve DCNN performance in detecting privacy threats such as unauthorized access, abnormal usage patterns, and data tampering. This optimization allows the model to identify complex intrusion patterns with greater precision. Python was used for model implementation and experimentation. Experimental results show that the SWO‐DCNN model achieves an overall performance accuracy of above 98% in detecting potential privacy breaches, outperforming conventional models. In addition, the proposed system supports real‐time monitoring and alert mechanisms, making it practical for integration into educational data infrastructures. This research highlights the potential of combining DL with intelligent optimization techniques to strengthen student data privacy in scalable and adaptive ways.