"Why can the private key still be used for decryption during ReLU computation? This seems to differ from the model description in the paper."

narger-ef / LowMemoryFHEResNet20

Source code for the paper "Encrypted Image Classification with Low Memory Footprint using Fully Homomorphic Encryption"

https://eprint.iacr.org/2024/460

MIT License

24 stars 7 forks source link

"Why can the private key still be used for decryption during ReLU computation? This seems to differ from the model description in the paper." #13

Closed yellow123Nike closed 2 months ago

yellow123Nike commented 2 months ago

narger-ef commented 2 months ago

Ops, it was added for empirically observe the range of values given as input to the approximation. Feel free to remove those lines, the decryption is not useful for the computation of the circuit. :-)

yellow123Nike commented 2 months ago

ok,thanks! This is great work. I'm slowly understanding based on the code and the paper.

narger-ef commented 2 months ago

No worries, thank you very much! Feel free to open the issue (or other issues) in case you are not fully convinced