English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Training even moderately-sized generative models with differentially-privatestochastic gradient descent (DP-SGD) is difficult: the required level of noisefor reasonable levels of privacy is simply too large. We advocate insteadbuilding off a good, relevant representation on an informative public dataset,then learning to model the private data with that representation. Inparticular, we minimize the maximum mean discrepancy (MMD) between privatetarget data and a generator's distribution, using a kernel based on perceptualfeatures learned from a public dataset. With the MMD, we can simply privatizethe data-dependent term once and for all, rather than introducing noise at eachstep of optimization as in DP-SGD. Our algorithm allows us to generateCIFAR10-level images with $\epsilon \approx 2$ which capture distinctivefeatures in the distribution, far surpassing the current state of the art,which mostly focuses on datasets such as MNIST and FashionMNIST at a large$\epsilon \approx 10$. Our work introduces simple yet powerful foundations forreducing the gap between private and non-private deep generative models. Ourcode is available at \url{https://github.com/ParkLabML/DP-MEPF}.

Pre-trained Perceptual Features Improve Differentially Private Image Generation