Preventing overfitting in deep learning using differential privacy

This research explores a differential-privacy based approach to improve generalization and prevent overfitting in Deep Neural Networks. Overfitting, where models learn noise and perform poorly on unseen data, is a growing challenge in modern AI systems.