This project is an attempt to create a simple machine learning library containing implemenations of a commonly used algorithms implemented in a way that is easy to understand.

Requirements

The requirements can be installed by running:

pip install -r requirements.txt

Useful guides:

• Peter Roelants' guide to Implementing neural networks
• Chris McCormick's guide to understanding PCA
• CS231n's section about kNN

Examples

Running the MNIST example

Begin by downloading the MNIST dataset from the MNIST database homepage. The location of the dataset is hardcoded to data in the file nn_mnist.py, but feel free to change it.

After having installed the requirements, you should be able to run python -m examples.nn_mnist, which trains the network and prints the final accuracy.