A Revolutionary Technique That Changed Machine Vision

Re:A Revolutionary Technique That Changed Machine Vision by kastnerkyle on Wednesday September 10, 2014 @07:31AM (#8296031)

Well, they have no choice. Because technically the copyright of each image is still held by the people who took the images (or in some cases the people in the images). Are weights of a trained network based on ImageNet a derived work, a cesspool of millions of copyright claims? What about networks trained to appoximate another ImageNet network?

There is a lot of legal gray area here unfortunately, so the signing of the license to work with the images seems like a CYA move to me.

[ Parent | on HN ]

Re:A Revolutionary Technique That Changed Machine Vision by conjectures on Wednesday September 10, 2014 @07:46AM (#8296074)

'Are weights of a trained network based on ImageNet a derived work, a cesspool of million of copyright claims?'

I'd argue not. The success of deep learning in vision seems to be in acquiring allocentric representations of objects. Copyright protects the expression and not the concept. Parameter weights describe the concept, not particular instantiations of it.

[ Parent | on HN ]

Re:A Revolutionary Technique That Changed Machine Vision by kastnerkyle on Wednesday September 10, 2014 @08:51AM (#8296290)

This is what I am hoping too - but derived works are a sticky subject. Imagine a scenario where someone gets of one of these "data vaults" from a large company, then trains a network and throws away the training data. You are still holding the "essence" of their datastore, even without the actual data.

I guess we won't really know until someone goes to court over it.

[ Parent | on HN ]

Re:A Revolutionary Technique That Changed Machine Vision by Houshalter on Thursday September 11, 2014 @06:12AM (#8301719)

You can't copyright "essence". There are weird legalities about derivative works, but this is an entirely different thing.

[ Parent | on HN ]