Skip to main content
. Author manuscript; available in PMC: 2021 Oct 13.
Published in final edited form as: J Mach Learn Res. 2021 Mar;22(141):1–49.

Figure 14:

Figure 14:

Cross-validation error in classifying the subject labels of the HNU1 data. For each embedding estimated with a different method, a 1-NN classifier was trained using the distance between the individual graph embeddings, varying the number of embedding dimensions (left plot) which control the description length of the different embeddings (right plot). MASE, OMNI and CISE are the only methods that achieve perfect accuracy, but MASE requires much fewer parameters.