Just so it's clear, are you claiming here that no deep learning is interpretable?
Starting May 25, the European Union will require algorithms to explain their output, making deep learning illegal.
-
-
-
(and just so I'm being clear, I think that would be a bit of a hyperbolic claim)
Keskustelun loppu
Uusi keskustelu -
-
-
Don't you think methods such as LIME can give acceptable explanations? For images, what about the various saliency method?
-
Hi
@tdietterich I would argue forcefully that LIME offers nothing of value in these critical cases. The local shape of the decision boundary at best gives practitioner intuition, but should never be used to substantiate a decision to grant bail. - Näytä vastaukset
Uusi keskustelu -
-
-
That kinda makes human intelligence--which is unexplainable--illegal by inference.
-
Humans can explain their behaviour. All creative behaviour is fallible.
Keskustelun loppu
Uusi keskustelu -
-
-
as you already said, "regulate first, shot later" ... but we need to think carefully about GDPR, https://research.google.com/pubs/pub45428.html …pic.twitter.com/17GI9B1xkM
Kiitos. Käytämme tätä aikajanasi parantamiseen. KumoaKumoa
-
-
-
that's a weird take on it. It will demand users to consent to their data to be use for whatever. You can still do deep learning on good data!
Kiitos. Käytämme tätä aikajanasi parantamiseen. KumoaKumoa
-
-
-
And how, exactly, are you supposed to validate the output?
Kiitos. Käytämme tätä aikajanasi parantamiseen. KumoaKumoa
-
-
-
It would suffice to give the training targets or in case of RL the reward function.
Kiitos. Käytämme tätä aikajanasi parantamiseen. KumoaKumoa
-
Lataaminen näyttää kestävän hetken.
Twitter saattaa olla ruuhkautunut tai ongelma on muuten hetkellinen. Yritä uudelleen tai käy Twitterin tilasivulla saadaksesi lisätietoja.