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

Launch HN: Flower (YC W23) – Train AI models on distributed or sensitive data

news.ycombinator.com

Post reply on HN
Print view View original

61–70 of 70 posts

Re: Launch HN: Flower (YC W23) – Train AI models on distributed or sensitive data

#61

Post by onethought »

aix1 wrote: post #57
onethought wrote: post #56

Earlier quoted context omitted.

This doesn’t sound right, if they don’t know the structure of the NN how can the reconstruct from the weights alone? (Perhaps the structure is communicated within the weights?)

Every agent training the model on their proprietary data has to have access to the model form in some way (otherwise how would they train it?) For this reason, one must assume that the model form is known to the adversary. With this, the question becomes: is it possible to reconstruct training data from a trained model? We already know that, at least for some image models, the answer to that question is "yes": https:…

That must only be true if there isn’t a one way compression step occurring, or any approximation in the whole model.

Re: Launch HN: Flower (YC W23) – Train AI models on distributed or sensitive data

#62

Post by cs02rm0 »

In the past, we funded our work through consulting projects, but looking ahead, we’re going to offer a managed version for enterprises and charge per deployment or federation.

Interesting.

Flower seems to fit well for people who are sensitive about their data and don't want to hand it over to a third party, but this seems to move towards a model where they have to hand that sensitive data over to a third party.

Perhaps that still works for the bulk of users, especially commercial rather than government. It's difficult to pursue both a managed solution and simultaneously maintain an open source offering without one departing from the other.

Re: Launch HN: Flower (YC W23) – Train AI models on distributed or sensitive data

#63

Post by aix1 »

onethought wrote: post #61
aix1 wrote: post #57

Earlier quoted context omitted.

Every agent training the model on their proprietary data has to have access to the model form in some way (otherwise how would they train it?) For this reason, one must assume that the model form is known to the adversary. With this, the question becomes: is it possible to reconstruct training data from a trained model? We already know that, at least for some image models, the answer to that question is "yes": https:…

That must only be true if there isn’t a one way compression step occurring, or any approximation in the whole model.

I don't think lossy compression is sufficient. The very first example in the paper I linked to is clearly not identical to the original image (=lossily compressed) yet leaks a training image in a way that would be highly problematic in certain domains, e.g. medical imaging.

Re: Launch HN: Flower (YC W23) – Train AI models on distributed or sensitive data

#64

Post by 1wheel »

dang wrote: post #53
niclane7 wrote: post #52

Earlier quoted context omitted.

It is reasonable to think of it that way. Certainly high-level information from the data is extracted and embedded within a model, but only the information necessary for the model being trained. Whereas if the data itself was being sent, then all of the information is available. Additionally, through added protections (differential privacy being one) it is possible to engineer the federated system such that the data…

Can you say more about what differential privacy is and how it works, for those of us who don't know or don't remember?

https://pair.withgoogle.com/explorables/private-and-fair/

Re: Launch HN: Flower (YC W23) – Train AI models on distributed or sensitive data

#65

Post by wjnc »

cs02rm0 wrote: post #62

In the past, we funded our work through consulting projects, but looking ahead, we’re going to offer a managed version for enterprises and charge per deployment or federation. Interesting. Flower seems to fit well for people who are sensitive about their data and don't want to hand it over to a third party, but this seems to move towards a model where they have to hand that sensitive data over to a third party. Perha…

Charge per deployment is on-prem? You bring the hardware, they send you the software.

Re: Launch HN: Flower (YC W23) – Train AI models on distributed or sensitive data

#66

Post by popinman322 »

cs02rm0 wrote: post #62

In the past, we funded our work through consulting projects, but looking ahead, we’re going to offer a managed version for enterprises and charge per deployment or federation. Interesting. Flower seems to fit well for people who are sensitive about their data and don't want to hand it over to a third party, but this seems to move towards a model where they have to hand that sensitive data over to a third party. Perha…

I didn't read it as a move towards centralizing data, but instead as working with companies to federate over their userbase or between a collection of companies.

Re: Launch HN: Flower (YC W23) – Train AI models on distributed or sensitive data

#67

Post by tanto »

wjnc wrote: post #65
cs02rm0 wrote: post #62

In the past, we funded our work through consulting projects, but looking ahead, we’re going to offer a managed version for enterprises and charge per deployment or federation. Interesting. Flower seems to fit well for people who are sensitive about their data and don't want to hand it over to a third party, but this seems to move towards a model where they have to hand that sensitive data over to a third party. Perha…

Charge per deployment is on-prem? You bring the hardware, they send you the software.

Hey wjnc, You can think of it how you can use GitLab on gitlab.com or deploy it yourself on-prem. The only difference being instead of per user we would charge per deployment. As in case of GitLab you can decide to host it yourself.

Re: Launch HN: Flower (YC W23) – Train AI models on distributed or sensitive data

#68

Post by onethought »

aix1 wrote: post #63
onethought wrote: post #61

Earlier quoted context omitted.

That must only be true if there isn’t a one way compression step occurring, or any approximation in the whole model.

I don't think lossy compression is sufficient. The very first example in the paper I linked to is clearly not identical to the original image (=lossily compressed) yet leaks a training image in a way that would be highly problematic in certain domains, e.g. medical imaging.

I see what you are saying. Agree. Seems we need some set patterns in NN models that will reliably remove reversibility without effecting loss too drastically.

Re: Launch HN: Flower (YC W23) – Train AI models on distributed or sensitive data

#69

Post by danieljanes »

fedgenerativeai wrote: post #30

[dead]

One of the creators of Flower here - I can only say that the team behind Flower appreciates the contributions of FedML to the field of federated learning. Their work helps to make federated learning more widely known, and they published significant advances in making federated learning more robust and scalable.

In fact, we are in the process of implementing LightSecAgg, and we'd welcome their feedback once we have a working version.

Re: Launch HN: Flower (YC W23) – Train AI models on distributed or sensitive data

#70

Post by jleguina »

This is a great project!

Have you though about what happens at inference? Suppose I train in a federated healthcare environment using PII features from patient records. Once I get the weights back how can I ever deploy it if I don't have access to the same features? The models would become highly coupled to the training environments no?

Best of luck!

Post reply on HN