Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Homomorphically encrypted CIFAR-10 inference in 200ms

sofar.belfortlabs.cloud

Post reply on HN
Print view View original

31–40 of 45 posts

Re: Homomorphically encrypted CIFAR-10 inference in 200ms

#31

Post by digitallawyer »

rcr-anti wrote: post #14

Looked at the network logs and the JS, did some testing, there's a caveat here. For an encryption demo you might expect your secrets to be generated locally, they do the compute on something they can't read, you compare their results to your original plaintext; (imo at least) the point would be that it isn't physically possible for them to cheat. Here, you literally download client_secret.bin from their server, so th…

I'm familiar with the team - for what it's worth, its a bunch of legit scientists out of KU Leuven (Belgium) not some vaporware crypto scam lol

Re: Homomorphically encrypted CIFAR-10 inference in 200ms

#33

Post by binary132 »

smalltorch wrote: post #7
furkanturan wrote: post #4

Earlier quoted context omitted.

Consider encrypted AI. You ask a question under encryption. The remote calculates the answer, which is still under encryption. This is the critical point; the remote cannot see the question and answer. It only calculates. Once you receive the answer, you decrypt it and see the answer; only you see the answer.

Ah, that sounds impossible. Good luck, sir...Also, who wants this if you can just run a local model?

maybe it could be running a much more powerful system than you have access to

Re: Homomorphically encrypted CIFAR-10 inference in 200ms

#34

Post by mswphd »

j2kun wrote: post #15
smalltorch wrote: post #12

Earlier quoted context omitted.

Have you considered that maybe the computation itself on the private data, no matter how you put it, even if encrypted , was designed to protect the consumer? I.E., gaining any sort of insight a transaction of protected information is what the protections were in place for. So is FHE more about skirting regulations and privacy laws? Or, is it a new frontier of an untapped data source that has some red tape around it?…

> gaining any sort of insight The server providing the FHE-based service does not gain any sort of insight. This is a key point: only the client can see the output of the computation.

this is only true in the IND-CPA model for most "practical" FHE work. So a more precise way to describe things is "a server who faithfully performs the task given to them does not gain any sort of insight" (perhaps with an additional caveat about decryption failures).

Re: Homomorphically encrypted CIFAR-10 inference in 200ms

#35

Post by deckar01 »

Eduard wrote: post #22
deckar01 wrote: post #16

I think they only trained on dogs with floppy ears, because it is very confident the German Shepherd is a cat.

did they remove a German Shepherd from the sample images since your comment? I cannot see a German Shepard. Only a black dog with floppy ears.

I uploaded my own image, tried various crops, but it always guessed cat or deer, with dog<3%.

Re: Homomorphically encrypted CIFAR-10 inference in 200ms

#36

Post by Pro7ech »

grg0 wrote: post #32

> Image classification without the server seeing the image. Does a given plain image always result in the same encrypted payload?

It uses randomized symmetric encryption with fresh public randomness for every encryption. Ciphertexts are computationally indistinguishable from random and do not reveal whether two encryptions correspond to the same plaintext, different plaintexts, the same key, or different keys.

Re: Homomorphically encrypted CIFAR-10 inference in 200ms

#37

Post by Pro7ech »

jszymborski wrote: post #5

Big if true! One of the caveats here is that the file size seems to balloon 341 times.

This is specific to the current state of the demo. Additional techniques can reduce fresh encryptions to only a few times the plaintext size, but they have not been deployed yet.

Re: Homomorphically encrypted CIFAR-10 inference in 200ms

#38

Post by Pro7ech »

smalltorch wrote: post #7
furkanturan wrote: post #4

Earlier quoted context omitted.

Consider encrypted AI. You ask a question under encryption. The remote calculates the answer, which is still under encryption. This is the critical point; the remote cannot see the question and answer. It only calculates. Once you receive the answer, you decrypt it and see the answer; only you see the answer.

Ah, that sounds impossible. Good luck, sir...Also, who wants this if you can just run a local model?

A local model works when one party owns everything. But often you'll have more than one party, for example different machine operators, the machine manufacturer, the sensor provider, all involved, each with data they cannot or IP they do not want share. That’s where FHE becomes truly useful and more than an additional layer of protection by enabling collaborations that were simply not possible before.

Re: Homomorphically encrypted CIFAR-10 inference in 200ms

#39

Post by Pro7ech »

mswphd wrote: post #34
j2kun wrote: post #15

Earlier quoted context omitted.

> gaining any sort of insight The server providing the FHE-based service does not gain any sort of insight. This is a key point: only the client can see the output of the computation.

this is only true in the IND-CPA model for most "practical" FHE work. So a more precise way to describe things is "a server who faithfully performs the task given to them does not gain any sort of insight" (perhaps with an additional caveat about decryption failures).

In the case of single key FHE with server-side processing (the demo setup), the server can never gains any insight, even as an active adversary that can arbitrarily deviate from the protocol because it only has access to public evaluation keys and ciphertexts. The worst it can do is return tempered output to the client, but it doesn't help it learning anything about the client request.
Post reply on HN