In the privacy-utility tradeoff of a model trained on benchmark language and vision tasks,
remarkable improvements have been widely reported when the model is pretrained on …

Diffusion models (DMs) are widely used for generating high-quality high-dimensional
images in a non-differentially private manner. To address this challenge, recent papers …

We show that the canonical approach for training differentially private GANs--updating the
discriminator with differentially private stochastic gradient descent (DPSGD)--can yield …

The availability of rich and vast data sources has greatly advanced machine learning
applications in various domains. However, data with privacy concerns comes with stringent …

Differential Privacy (DP) image data synthesis, which leverages the DP technique to
generate synthetic data to replace the sensitive data, allowing organizations to share and …

Data sharing has led to the emergence of the deep generative model (DGM) with differential
privacy for synthesizing tabular data. However, existing methods struggle to synthesize high …

Recent advances in score-based generative models have led to a huge spike in the
development of downstream applications using generative models ranging from data …

The digital era is characterized by the widespread availability of rich data, which has fueled
the growth of machine learning applications across diverse fields such as computer vision …

Data privacy protection is garnering increased attention among researchers. Diffusion
models (DMs), particularly with strict differential privacy, can potentially produce images with …

Recent advances in score-based generative models have led to a huge spike in the
development of downstream applications using generative models ranging from data …