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

Machine Unlearning in 2024

ai.stanford.edu

Post reply on HN
Print view View original

71–80 of 97 posts

Re: Machine Unlearning in 2024

#71

Post by eru »

avi_vallarapu wrote: post #44
isodev wrote: post #41

Earlier quoted context omitted.

> a time-to-unlearn kind of an acceptable agreement Why put the burden to end users? I think the technology should allow for unlearning and even "never learn about me in any future models and derivative models".

No technology can guarantee 100% unlearning, and the only 100% guarantee is when the data is deleted before the model is retrained. Legally, even 99.99% accuracy may not be acceptable, but, only 100%.

Or rather some legal fiction that you can pretend is 100%. You can never achieve real 100% in practice after all. Eg the random initialisation of weights might already encode all the 'bad' stuff you don't want. Extremely unlikely, but not strictly 0% unlikely.

The law cuts off at some point, and declares it 100%.

Re: Machine Unlearning in 2024

#72

Post by mr_toad »

avi_vallarapu wrote: post #44
isodev wrote: post #41

Earlier quoted context omitted.

> a time-to-unlearn kind of an acceptable agreement Why put the burden to end users? I think the technology should allow for unlearning and even "never learn about me in any future models and derivative models".

No technology can guarantee 100% unlearning, and the only 100% guarantee is when the data is deleted before the model is retrained. Legally, even 99.99% accuracy may not be acceptable, but, only 100%.

> the only 100% guarantee is when the data is deleted before the model is retrained

That’s not even a guarantee. A model can hallucinate information about anyone, and by sheer luck some of those hallucinations will be correct. And as a consequence of forging (see section 2.2.1) you’d never be able to prove whether the data was in the training set or not.

Re: Machine Unlearning in 2024

#73

Post by mr_toad »

nullc wrote: post #10

I've wondered before if it was possible to unlearn facts, but retain the general "reasoning" capability that came from being trained on the facts, then dimensionality reduce the model.

How much reasoning capability LLM’s have is up for debate.

With a true AGI you could just tell it to keep people’s personal information confidential and expect that it would understand that instruction.

Re: Machine Unlearning in 2024

#74

Post by ekianjo »

beeboobaa3 wrote: post #60
cscurmudgeon wrote: post #58

Earlier quoted context omitted.

1. You are assuming just training a model on copyrighted material is a violation. It is not. It may be under certain conditions but not by default. 2. Why should we aim for harsh punitive punishments just because it was done so in the past?

> 1. You are assuming just training a model on copyrighted material is a violation. It is not. It may be under certain conditions but not by default. Using copyrighted content for commercial purposes should be a violation if it's not already considered to be one. No different from playing copyrighted songs in your restaurant without paying a licensing fee. > 2. Why should we aim for harsh punitive punishments just be…

> Using copyrighted content for commercial purposes should be a violation

so reading a book and using the book contents to help you in your job would be a violation too based on your logic

Re: Machine Unlearning in 2024

#75

Post by beeboobaa3 »

ekianjo wrote: post #74
beeboobaa3 wrote: post #60

Earlier quoted context omitted.

> 1. You are assuming just training a model on copyrighted material is a violation. It is not. It may be under certain conditions but not by default. Using copyrighted content for commercial purposes should be a violation if it's not already considered to be one. No different from playing copyrighted songs in your restaurant without paying a licensing fee. > 2. Why should we aim for harsh punitive punishments just be…

> Using copyrighted content for commercial purposes should be a violation so reading a book and using the book contents to help you in your job would be a violation too based on your logic

A business cannot read a book, and your machine learning model is not given human rights.

Re: Machine Unlearning in 2024

#77

Post by isodev »

eru wrote: post #71
avi_vallarapu wrote: post #44

Earlier quoted context omitted.

No technology can guarantee 100% unlearning, and the only 100% guarantee is when the data is deleted before the model is retrained. Legally, even 99.99% accuracy may not be acceptable, but, only 100%.

Or rather some legal fiction that you can pretend is 100%. You can never achieve real 100% in practice after all. Eg the random initialisation of weights might already encode all the 'bad' stuff you don't want. Extremely unlikely, but not strictly 0% unlikely. The law cuts off at some point, and declares it 100%.

All this is technically correct, but it also means this technology is absolutely not ready to be used for anything remotely involving humans or end user data.

Re: Machine Unlearning in 2024

#78

Post by Dylan16807 »

beeboobaa3 wrote: post #75
ekianjo wrote: post #74

Earlier quoted context omitted.

> Using copyrighted content for commercial purposes should be a violation so reading a book and using the book contents to help you in your job would be a violation too based on your logic

A business cannot read a book, and your machine learning model is not given human rights.

> A business cannot read a book

Assume the human read the book as part of their job. Is that using copyrighted material for commercial purposes?

If that doesn't count then I'm not sure why you brought up "commercial purposes" at all.

> This rules with harsh penalties for consumers/small companies but not for bigtech double standard is bullshit, though.

Consumers and small companies get away with small copyright violations all the time. And still bigger than having your image be one of millions in a training set.

Re: Machine Unlearning in 2024

#80

Post by eru »

isodev wrote: post #77
eru wrote: post #71

Earlier quoted context omitted.

Or rather some legal fiction that you can pretend is 100%. You can never achieve real 100% in practice after all. Eg the random initialisation of weights might already encode all the 'bad' stuff you don't want. Extremely unlikely, but not strictly 0% unlikely. The law cuts off at some point, and declares it 100%.

All this is technically correct, but it also means this technology is absolutely not ready to be used for anything remotely involving humans or end user data.

Why? We use random data in lots of applications, and there's always the theoretical probability that it could 'spell something naughty'.
Post reply on HN