Swift: Google’s Bet on Differentiable Programming

Re:Swift: Google’s Bet on Differentiable Programming by downerending on Thursday April 09, 2020 @01:06PM (#22824713)

Is the presumption that these parameters are floats, or at least numeric? It seems like if I took the functions from some random program (e.g., GNU diff), most would not be meaningfully differentiable. Or perhaps I'm missing something?

[ Parent | on HN ]

Re:Swift: Google’s Bet on Differentiable Programming by ummonk on Thursday April 09, 2020 @05:35PM (#22827461)

Yes, the parameters you're differentiating with respect to need to be floats.

Though there might be potential for extending the frameworks to e.g. differential cryptanalysis - I'm not knowledgeable enough about it to say how much differential cryptanalysis can be done programmatically.

[ Parent | on HN ]

Re:Swift: Google’s Bet on Differentiable Programming by lalord on Thursday April 09, 2020 @01:56PM (#22825321)

Yeah, they have to be semantically continuous, ie floats but not ints.

[ Parent | on HN ]