Probabilistic Quantum Clustering (PQC)

Usage

• PQC_main_script.m contains the main script with the pipeline of the PQC algorithm.

-Before, add project folder to Matlab search path.

• The script uses 5 datasets as examples, they are contained in datasets4.mat, where one of them is selected by the opt variable.

• There are two model variants implemented, QC2 and QC3. They can be switched with QC3 = True/False

  • QC2 is more basic and estimates the density of the points by K-NN with spherical distributions.
  • QC3 is more complex and estimates the density of the points by K-NN with distributions based on covariance matrices.

• The two main hyper-parameters can be scanned enabling the following variables (scan_knn and scan_dE). The graph with average negative log-likelihood (ALL) will help to select the most appropriate hyper-parameter (the smaller ALL the better).

TODO: Add additional guidelines to clarify the usage

Additional information and mathematical background can be found in published articles:

• Current version in Knowledge-Based Systems

• arXiv preliminar version