General ML

These tutorials introduce simple machine-learning examples primarely using neural networks with familiar datasets typically used for pedagogical purposes, keeping the focus on implementation.

Topics

• Software Toolkit: a quick map of the Python packages and data structures used in the examples.

• Neural Networks and Convolutional Neural Networks: complete MNIST workflows using native PyTorch and Keras on the Torch backend.

• CIFAR-10 CNN Progression, Transfer Learning, and Generative Models: progressively more involved vision examples.

• Tree Models and Ensembles and Hyperparameter Tuning: boosted trees and some examples for hyperparameter tuning.

Native PyTorch is the canonical neural-network implementation. The shorter Keras 3 notebooks use the same Torch runtime and show the equivalent high-level compile() and fit() workflow that one may prefer to use for simplicity. The notebooks also include ideas to experiment interactively.