RNNs: http://karpathy.github.io/2015/05/21/rnn-effectiveness/
https://hackage.haskell.org/package/grenade
haskell, composing well FP, succintness is important for complex concepts involved in RNN, and using an RNN as another layer in Grenade is very elegant and this is enabled by the haskell's composability, type-safety and type-classes. This makes RNNs almost fun relative to how painful tensorflow and their imperative kin are to use in python. Albeit tensorflow is more advanced with more advanced features not yet available in weka, grenade, hlearn.
not all neural networks are equal imho: https://www.youtube.com/watch?v=sPjA6lS0GlQ
>> what’s really nice about these functions is that they directly mirror the functions in Layers, in fact, this match works so well that we can embed a Network as a Layer in another Network. Grenade is a dependently typed, high level, artificial neural network library written in Haskell. It’s pretty robust, fast, and expressive, and uses some interesting type level programming techniques which don’t appear to have been used in neural networks before. Grenade does a lot of things right, it’s very expressive in my view, and makes the construction of networks incredibly safe: they essentially can’t crash, and they’ll always be sound. It’s also pretty quick, using BLAS and some hand written C, I’ve found this acceptable for a lot of problems.
or for lispers : http://www.intelligency.org/ai_lisp.php
or for elixir : https://dashbit.co/blog/elixir-and-machine-learning-nx-v0.1