Skip to main content
. 2020 Oct 30;6(44):eabd0855. doi: 10.1126/sciadv.abd0855

Fig. 5. Comparison of DCA-based scLearn and MDDM-based scLearn for multilabel single-cell assignment.

Fig. 5

(A) Geometric intuition for DCA-based learning and MDDM-based learning. (B) Visualization of clustering results by UMAP before and after MDDM-based transforming for the ESC (embryonic stem cells) dataset. (C) The micro_F1 of DCA-based scLearn and MDDM-based scLearn under two strategies for three multilabel datasets were presented. mEM, mouse embryo ventral midbrain cells; hPFC, human prefrontal cortex cells.