Tutorials

Interactive Jupyter notebook tutorials demonstrating various features of TorchGMM.

Getting Started

• Gaussian Mixture Models (GMM) - Basic introduction to fitting GMMs with TorchGMM, including a comparison of mean/weight/covariance initialization strategies
• EM Algorithm Walkthrough - Step-by-step visualization of how the EM algorithm updates responsibilities and parameters each iteration

Advanced Topics

• Classification EM (CEM) - Comparing the hard-assignment CEM algorithm against standard EM
• Prior Distributions - Using priors for regularization and MAP estimation
• NIW Priors Comparison - Detailed comparison of Normal-Inverse-Wishart priors
• Sampling from GMMs - Generating synthetic data from fitted models

Visualization

• PCA Plotting - Visualizing high-dimensional GMM results with PCA
• Visualization Techniques - Comprehensive guide to plotting GMM results

Model Evaluation

• Clustering Metrics - Evaluating GMM performance with various metrics

Running Tutorials

Option 1: Google Colab

Click the "Open in Colab" button at the top of each notebook.

Option 2: Local Jupyter

Install Jupyter:

pip install jupyter

Launch Jupyter:

cd docs/notebooks
jupyter notebook

Option 3: JupyterLab

pip install jupyterlab
jupyter lab

Requirements

All tutorials require:

pip install torch numpy matplotlib scipy
pip install tgmm  # or install from source

Some tutorials may have additional dependencies listed at the top of the notebook.