Fig. 3

Illustration of manifold embedding and clustering on simulated Gaussian curves. a Four groups of parameters including two ranges of mean in [0, 1], [10,11] with the step size of 0.01, and two standard deviations of 0.2, 1. b Four groups of Gaussian curves, generated from the four groups of parameters in a. Each Gaussian curve has a unique combination of mean and standard deviation. c–e Manifold points estimated by Isomap⁵¹, colored by true labels and clustered labels via K-means and hierarchical density estimates method (HDBSCAN⁵⁷), respectively. f–h Manifold points estimated by t-SNE⁵⁵, colored by true labels and clustered labels via K-means and HDBSCAN, respectively. i–k Manifold points estimated by Graph-Bootstrapping (GB), colored by true labels and clustered labels via K-means and HDBSCAN, respectively