Artistic expression and data protection: Balancing aesthetics with data privacy in IoT

Abstract

Integrating Internet of Things (IoT) technologies in art design has created new possibilities for artists to create immersive and interactive experiences. However, data collection, analysis, and utilization in IoT art installations raise significant security and privacy concerns. Additionally, incorporating differential privacy techniques in IoT art installations poses optimization challenges. This paper explores optimizing differential privacy budgets based on deep learning in IoT art installations. By leveraging deep learning models, privacy budgets can be dynamically allocated to preserve individual privacy while maintaining the aesthetic integrity of the artwork. In light of this, a deep learning-based differential privacy budget optimization strategy for IoT art installations is suggested. This method adaptively distributes various budgets by the iterative change law of parameters. A regularization term is provided to limit the disturbance term to avoid the issue of excessive noise. This stops the neural network from overfitting and also assists in learning the model's salient characteristics. The capacity of the model to generalize is effectively improved by the suggested strategy, according to experiments. The accuracy difference between the model trained with noise and the model trained with original data is less than 0.5% as the number of iterations increases. Therefore, the proposed method can protect the user's privacy, effectively ensure the model's availability, and achieve the balance between privacy and availability. This accuracy ensures that the installation functions as intended and delivers the desired aesthetic impact, enabling artists to convey their artistic message effectively.