Aloïs Bissuel

I guess supporting gaussian noise instead of a laplacian one would improve the utility of the API when reporting over a few dimensions.

Sorry for the very late answer. It looks like this mechanism should enable most of our use cases. A quick question though: would a source have only one dedup_key associated...

Thanks for the quick answer, and sorry for not following it up right away. I am not sure I follow the second solution. Isn't it the very role of priorities...

Have you considered enabling the training of ML models directly in the TEE? This would create a lot of value for the API, as it is arguably a lot easier...

Thanks for you answer. Regarding the two concerns you voiced, may I ask some further clarifications? 1. Performance. Could you confirm that the performance hit is due to MPC? 2....

2. Got it! This definitely makes sense. Here, Linf is used to control the L2 sensitivity for Gaussian noise we applied. In other words, if Linf=1, the L2 sensitivity is...

I am not sure to follow your proposal. What do you mean by parties and users here?

This would be definitively helpful to enable users of the API to get more effective measurement without having to introduce too much complexity on the server-side. Not sure if this...

Thanks for the quick answer! Using a dense encoding for the publisher is challenging for a number of reasons: - the list of publishers is very dynamic. As a rule...

Thanks for the detailed answer. It looks like it would indeed solve our use case, depending on the default permission being discussed in #393.