GNODEVAE: a graph-based ODE-VAE enhances clustering for single-cell data.

ularization, and for reconstruction quality metrics, GNODEVAE achieved statistically significant improvements in ARI and NMI compared to both scGCC and scGNN, demonstrating the advantages of generative modeling in maintaining data reconstruction fidelity, while differences with CLEAR in ARI and NMI did not reach statistical significance, reflecting the competitiveness of contrastive learning methods in certain reconstruction-related tasks (Fig. 6 A, Table 14 ).

also in 111,027 other papers