Privacy-preserving machine learning aims to train models on private data...
We explore Reconstruction Robustness (ReRo), which was recently proposed...
The ability to generate privacy-preserving synthetic versions of sensiti...
Auditing mechanisms for differential privacy use probabilistic means to
...
Image diffusion models such as DALL-E 2, Imagen, and Stable Diffusion ha...
Differential Privacy (DP) provides a formal privacy guarantee preventing...
Given access to a machine learning model, can an adversary reconstruct t...
We study the difficulties in learning that arise from robust and
differe...
Advancements in deep generative models have made it possible to synthesi...
The widespread adoption of encrypted communications (e.g., the TLS proto...
Federated Learning (FL) allows multiple participants to collaboratively ...
Historically, machine learning methods have not been designed with secur...
Randomized smoothing, a method to certify a classifier's decision on an ...
Machine learning models are vulnerable to adversarial perturbations, tha...
Machine learning is data hungry; the more data a model has access to in
...
Since Biggio et al. (2013) and Szegedy et al. (2013) first drew attentio...
Security-critical applications such as malware, fraud, or spam detection...
Neural networks are known to be vulnerable to adversarial examples, inpu...
Adversarial training was recently shown to be competitive against superv...