Closed siamakz closed 3 years ago
Question: When doing convolution opreation, does the pixels of the image are encrypted separately using LWE ciphertext ?
Or multiple pixels are packed into a polynomial before encryption ?
Hello :hand:
Thanks for your questions!
Concerning @siamakz question: we will provide a demo with a neural networks inference that uses concrete soon. We will ping you on the issue when the tutorial is released.
@fionser, we do not provide an homomorphic convolution operator in concrete as of now. If you want to implement it on your own, the simplest way is to encrypt each pixel as an LWE ciphertext, and operate on those.
Thank you for this revolutionary HE effort. I was wondering whether there is any end to end tutorial for implementing a neural network using Concrete?