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

A JPEG Image Compression Service Using Part Homomorphic Encryption (2019)

github.com

Post reply on HN
Print view View original

31–40 of 41 posts

Re: A JPEG Image Compression Service Using Part Homomorphic Encryption (2019)

#31

Post by snowfoolin »

magicalhippo wrote: post #29
snowfoolin wrote: post #25

Earlier quoted context omitted.

These are good questions! FHE appears nonintuitive at first glance, but hopefully with a smaller example it can be made clear. > Wouldn't the information required to identify what a "word" is require the running software to be able to see breaks/periods/etc? Yes. But that information can be encrypted and still computed on. You might suppose that since encrypted data is essentially gibberish, then multiplying or addin…

Great explanation, thank you. In your example you decryped by key^2. The decryption requiring a different or derivative key a general feature of the algorithm or just a quirk of your example?

In the ElGamal example I linked you'll see that we use the same key for encryption and decryption. However, we perform a modification on the cyphertext using known values. This has a similar effect.

It's also somewhat definitional. We're taking an established encryption scheme in both cases and constructing a new algorithm on top of it. Above where I say we're performing decryption with a key set to key^2, we could also think of it as doing Dec(key, key*m) where the key is the same but we've done some operation on the cyphertext.

Re: A JPEG Image Compression Service Using Part Homomorphic Encryption (2019)

#32

Post by stayclassyhn »

crazygringo wrote: post #19

Can the submitter, or anyone else, please explain what this is? There is zero information on the submitted page besides what's in the title, and this is a six-year-old GitHub repository with no activity in years. What are we looking at? What is the significance of this repo? Did it invent this idea? Does it have any practical uses? Is there any broader context at all?

Literally got flagged off this comment thread for saying the same thing...

Re: A JPEG Image Compression Service Using Part Homomorphic Encryption (2019)

#33

Post by notpushkin »

stayclassyhn wrote: post #32
crazygringo wrote: post #19

Can the submitter, or anyone else, please explain what this is? There is zero information on the submitted page besides what's in the title, and this is a six-year-old GitHub repository with no activity in years. What are we looking at? What is the significance of this repo? Did it invent this idea? Does it have any practical uses? Is there any broader context at all?

Literally got flagged off this comment thread for saying the same thing...

I think you’re talking about the comment by whatdoinghere (as I don’t see any dead comments in your name in this thread).

If that is your comment indeed, I think the problem was that you were overly dismissive and sarcastic. The parent poster asks questions instead, and doesn’t make remarks beyond stating the facts – this reads as harsh, but fair. (I also suppose they’ve accumulated some karma and this helps with the flagging algorithm somehow.)

Re: A JPEG Image Compression Service Using Part Homomorphic Encryption (2019)

#34

Post by magicalhippo »

snowfoolin wrote: post #31
magicalhippo wrote: post #29

Earlier quoted context omitted.

Great explanation, thank you. In your example you decryped by key^2. The decryption requiring a different or derivative key a general feature of the algorithm or just a quirk of your example?

In the ElGamal example I linked you'll see that we use the same key for encryption and decryption. However, we perform a modification on the cyphertext using known values. This has a similar effect. It's also somewhat definitional. We're taking an established encryption scheme in both cases and constructing a new algorithm on top of it. Above where I say we're performing decryption with a key set to key^2, we could a…

> In the ElGamal example I linked you'll see

Was on mobile phone so was too hard to read the handwritten text. Will definitely check it out later.

Anyway thanks again for the additional explanation, much appreciated.

Re: A JPEG Image Compression Service Using Part Homomorphic Encryption (2019)

#35

Post by totetsu »

xyzzyz wrote: post #22

If you are curious about what use case there is of it: imagine that instead of JPEG compression, we are running an LLM. You send your prompt encrypted, server homomorphically executes the model and sends the encrypted output, which you then decrypt. As a result, you get LLM output, but the server operators never learn anything about the contents of the query or response.

But.. if the LLM uses something like a MOE architecture, then surely some information about the prompt would leak just by observing what parts of the model are pulled into memory?

Re: A JPEG Image Compression Service Using Part Homomorphic Encryption (2019)

#36

Post by stayclassyhn2 »

notpushkin wrote: post #33
stayclassyhn wrote: post #32

Earlier quoted context omitted.

Literally got flagged off this comment thread for saying the same thing...

I think you’re talking about the comment by whatdoinghere (as I don’t see any dead comments in your name in this thread). If that is your comment indeed, I think the problem was that you were overly dismissive and sarcastic. The parent poster asks questions instead, and doesn’t make remarks beyond stating the facts – this reads as harsh, but fair. (I also suppose they’ve accumulated some karma and this helps with the…

"I think the problem was that you were overly dismissive and sarcastic."

Maybe.

"There is zero information on the submitted page besides what's in the title, and this is a six-year-old GitHub repository with no activity in years."

We made the same points though I'm admittedly more flippant and curt.

Re: A JPEG Image Compression Service Using Part Homomorphic Encryption (2019)

#37

Post by notpushkin »

stayclassyhn2 wrote: post #36
notpushkin wrote: post #33

Earlier quoted context omitted.

I think you’re talking about the comment by whatdoinghere (as I don’t see any dead comments in your name in this thread). If that is your comment indeed, I think the problem was that you were overly dismissive and sarcastic. The parent poster asks questions instead, and doesn’t make remarks beyond stating the facts – this reads as harsh, but fair. (I also suppose they’ve accumulated some karma and this helps with the…

"I think the problem was that you were overly dismissive and sarcastic." Maybe. "There is zero information on the submitted page besides what's in the title, and this is a six-year-old GitHub repository with no activity in years." We made the same points though I'm admittedly more flippant and curt.

Yeah, I agree, the point is basically the same – I think it’s the tone that was the problem. You don’t have to be overly positive and wrap your opinion with tons of fluff, just remember it’s other people’s work you’re talking about (and not everybody is good at explaining what they do!)

Re: A JPEG Image Compression Service Using Part Homomorphic Encryption (2019)

#38

Post by palata »

totetsu wrote: post #35
xyzzyz wrote: post #22

If you are curious about what use case there is of it: imagine that instead of JPEG compression, we are running an LLM. You send your prompt encrypted, server homomorphically executes the model and sends the encrypted output, which you then decrypt. As a result, you get LLM output, but the server operators never learn anything about the contents of the query or response.

But.. if the LLM uses something like a MOE architecture, then surely some information about the prompt would leak just by observing what parts of the model are pulled into memory?

I think if you want full homomorphic encryption, you have to go for the worst case. So you always pull all the model? That's usually why FHE is not practical.

Re: A JPEG Image Compression Service Using Part Homomorphic Encryption (2019)

#39

Post by nailer »

crazygringo wrote: post #19

Can the submitter, or anyone else, please explain what this is? There is zero information on the submitted page besides what's in the title, and this is a six-year-old GitHub repository with no activity in years. What are we looking at? What is the significance of this repo? Did it invent this idea? Does it have any practical uses? Is there any broader context at all?

The best reply is further down the thread, so I’ll just link to it:

https://news.ycombinator.com/item?id=43284843

Re: A JPEG Image Compression Service Using Part Homomorphic Encryption (2019)

#40

Post by daniellaigz »

thephyber wrote: post #30
crazygringo wrote: post #19

Can the submitter, or anyone else, please explain what this is? There is zero information on the submitted page besides what's in the title, and this is a six-year-old GitHub repository with no activity in years. What are we looking at? What is the significance of this repo? Did it invent this idea? Does it have any practical uses? Is there any broader context at all?

Not the poster or repo author, but: Homomorphic encryption is a way of applying logic transformations on encrypted data without having to decrypt it first. A simple example would be multiplying a number field in an encrypted blob by 2. The repo title is JPEG compression (a file format used for photograph-like images that uses lossy compression to reduce filesize). The repo links to another project which converts bitm…

Pls take a look at Lattica’s FHE survey, it shouldn't take more than ten minutes to complete and we’d appreciate your response https://forms.gle/XfKBUmDwpVHZkeTv7. Thank you!
Post reply on HN