pull down to refresh
21 sats \ 1 reply \ @OriginalSize 5 Feb \ on: Zero-Knowledge Proofs of Training for Deep Neural Networks crypto
How can you know that a NN was trained correctly without having trained it already? This feels like the halting problem.
That's the point of ZKPs. The prover trains it and creates the proof, then anybody can verify that it happened without revealing any more information.
reply