A functional approach to automatic differentiation

[darrenjw]

I could possibly do a session on this, but so could Jonny Law

[paulcc]

any useful references on this?

[jonnylaw]

A haskell implementation is here: http://hackage.haskell.org/package/ad
Here's a paper with an accompanying talk for a different haskell implementation: http://conal.net/papers/essence-of-ad/
Here's another implementation using continuation passing style, this time in Scala with an accompanying paper: https://github.com/feiwang3311/Lantern/

[darrenjw]

This is really interesting:
http://conal.net/papers/essence-of-ad/
but I don't yet understand it all...
See also:
https://stackoverflow.com/questions/2744973/is-there-any-working-implementation-of-reverse-mode-automatic-differentiation-fo

I was thinking about something more along the lines of:
https://skillsmatter.com/skillscasts/12818-deep-learning-programming-with-a-difference
https://github.com/noelwelsh/fdl

[darrenjw]

Great minds think alike...

[paulcc]

cheers both!