/MBMM-and-FBBMM

Primary LanguagePython

MBMM-and-FBBMM

  • Multivariate Beta Mixture Model (MBMM)
  • Flexible Bivariate Beta Mixture Model (FBBMM)

Setup

Tested under Python 3.10.4 in Ubuntu. Install the required packages by

$ pip install -r requirements.txt

Datasets

The following files are under folder data/

  • wine_2d.csv: Data from sklearn.datasets.load_wine() is reduced to 2-dimensions by AutoEncoder. Shape: (178, 2)
  • breast_cancer_2d.csv: Data from sklearn.datasets.load_breast_cancer() is reduced to 2-dimensions by AutoEncoder. Shape: (569, 2)
  • mnist_2d.csv: Data from MNIST dataset is reduced to 2-dimensions by CNN and AutoEncoder. Shape: (70000, 2) (including train and test data)

All AutoEncoder codes are also in the folder.

Execution

Compare k-means, MeanShift, DBSCAN, Agglomerative Clustering, GMM, MBMM and FBBMM on different datasets.

It is normal for FBBMM training time to be longer.

Synthetic dataset

  • Comparing different clustering algorithms on toy datasets. Reference from scikit-learn.
$ python3 synthetic.py

Wine dataset

  • Comparing different clustering except FBBMM on original features dataset.
  • Comparing different clustering on 2-dimensions features dataset.
  • Clustering performance evaluation metrics: Accuracy, ARI(Adjusted Rand Index), AMI(Adjusted Mutual Information).
$ python3 wine.py

Breast cancer dataset

  • Comparing different clustering except FBBMM on original features dataset.
  • Comparing different clustering on 2-dimensions features dataset.
  • Clustering performance evaluation metrics: Accuracy, ARI(Adjusted Rand Index), AMI(Adjusted Mutual Information).
$ python3 breast_cancer.py

MNIST dataset

  • Comparing different clustering on 2-dimensions feature dataset. Two Experinments: number0 and number3, number1 and number9
  • Clustering performance evaluation metrics: Accuracy, ARI(Adjusted Rand Index), AMI(Adjusted Mutual Information).
$ python3 MNIST.py