Open Access

Table B.1

Overview of the selected clustering algorithms, their hyperparameters, and their clustering types.

Algorithm Hyperparameters Type
KMeans init: [’k-means++’, ’random’], max_iter: [30000], tol: [1e-5] Hard
GMM covariance_type: [’full’, ’diag’, ’tied’, ’spherical’], tol: [1e-5], max_iter: [30000] Soft
Bayesian GMM covariance_type: [’full’, ’diag’, ’tied’, ’spherical’], tol: [1e-5], max_iter: [30000] Soft
Agglomerative linkage: [’ward’, ’complete’, ’average’, ’single’] Hard
BIRCH threshold: np.linspace(0.1, 0.5, 41) Hard
Spectral Clustering affinity: [’nearest_neighbors’, ’rbf’] Hard
DBSCAN eps: np.linspace(0.1, 2, 20), min_samples: [5, 10, 15] Hard
OPTICS min_samples: [5,10,15], xi: [0.05, 0.1, 0.2], min_cluster_size: [0.05, 0.1, 0.2] Hard

Notes. The notation np.linspace(start, stop, num) indicates an array of num evenly spaced numbers starting with start and ending with stop.

Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.

Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.

Initial download of the metrics may take a while.