Toy Machine Learning with Haskell
In this post, I show how you can use Haskell and the ad package (automatic differentiation) to build a toy machine learning model in Haskell. I’ve tried to write enough that someone without a machine learning background can follow the code. If you do have an ML background, just skim those parts!
I’m going to try something different. Instead of writing in Medium, I’ve written up this post in comments inside of CodeWorld.
Part 1: Model Structure
In this part, I explain what a machine learning model is, define the Haskell types for the parts of the model, and write the code to make the model work. We end with a visualization of a pre-trained model to recognize which points are in a circle.
Link: https://code.world/haskell#PisFfAHxvUXzYBjnDoeRmew
Part 2: Training
In this part, I show how to use automatic differentiation to train the model that we defined in the previous section.
