On Mon, Dec 28, 2009 at 05:14:53PM +0000, Amy de Buitléir wrote:
I'm building a library of components for artificial neural networks. I'm used to object-oriented languages, so I'm struggling a bit to figure out how to do a good design in a functional programming language like Haskell.
Q1: I've come up with two designs, and would appreciate any advice on improvements and what approach to take.
===== Design #1 ===== class Neuron n where activate :: [Double] -> n -> Double train :: [Double] -> Double -> n -> n
Looks reasonable.
The disadvantage of this approach is that I need to define and name each instance of neuron before I can use it. I'd rather create a neuron on-the-fly by calling a general-purpose constructor and telling it what functions to use for activation and training.
Indeed, if you want to create new types of neurons on-the-fly you should use your second design.
===== Design #2 ===== data Neuron = Neuron { weights :: [Double], activate :: [Double] -> Double, train :: [Double] -> Double -> Neuron }
Mostly makes sense. Note that this is really making explicit what the type class mechanism is doing -- a type class instance corresponds to a "dictionary", a record of type class methods, which is implcitly passed around. Here you are just explicitly declaring a dictionary type for neurons. The one thing that confuses me is why you included "weights" in the explicit dictionary but not in the Neuron type class. I would think you'd want it in both or neither, I don't see any reason you'd need it in one but not the other.
Q2: I thought there might be some way to define a function type, but the following doesn't work. Is there something along these lines that would work?
type activationFunction = [Double] -> Double
Type names must start with an uppercase letter. type ActivationFunction = [Double] -> Double should work just fine. -Brent