Spectral Clustering with MatLab

This is an intuitive implementation of Spectral Clustering with MATLAB. You can easily finish a spectral clustering analysis using Scikit-Learn similar API (the comparison between spectral clutsering and other clustering methods can be found here). For more deatils about spectiral clustering, you can read the references below or a brief introduction written by us.

Getting Started

1. Clone this repo and run main.m to see the demo
2. Usage Details

• Specify your data matrix X and label vector y, or get the built-in data

  [X, y] =  make_digits_dataset(300, true, false); % 300 examples, balance, no shuffle
  

  In this project, we provide 6 datasets including 3 toys and 3 real datasets. You can find them in datasets.

• Construct your graph

  W = fullyconnected(X, 7.8, 'rbf');
  

  There havs 3 types of graph you can use. Find them in graph.

• Run the Spectral Clustering Solver

  [C, ~] = SpectralClustering(W, size(unique(y), 1), 2);
  

  C is the predicted clusters.

• You can also easily metric the predicted answer with the functions under the metric.

Contributors

• Junhao Liu
• Jinlai Nin
• Dedong Huang
• Hao Wen

Reference

1. Von Luxburg, Ulrike. "A tutorial on spectral clustering." Statistics and computing 17.4 (2007): 395-416.