About this document

Privacy Preserving by eng23ra0043 is a document available to read on EtoBox.

The document presents a study on privacy-preserving federated learning (FL) that integrates adaptive noise scaling and enhanced convolutional neural network (CNN) models. It proposes a mechanism that adjusts noise based on client-specific loss variance to balance privacy and model utility, achieving high accuracy on the MNIST dataset while maintaining privacy. The findings suggest that adaptive differential privacy can effectively enhance federated learning applications in sensitive domains such as healthca

Author

eng23ra0043

Language

EN