bayesian-machine-learning

Why we included this project

The notebooks here work as a real curriculum rather than a pile of code snippets. Each topic starts from the math and shows a plain NumPy/SciPy implementation before, in several cases, redoing the same model in scikit-learn, PyMC3, GPy, or JAX, so you can see how theory becomes working code and how different frameworks express the same idea. Coverage spans Bayesian linear regression, Gaussian process regression and classification, sparse GPs, Bayesian optimization for hyperparameter tuning, variational inference, and variational autoencoders. That makes it a useful reference both for learners and for practitioners who hit a probabilistic model they do not fully trust. Notebooks open directly in Colab and formulas render cleanly via nbviewer, which makes working through the derivations at your own pace straightforward before you move to heavier production tooling.