Fig. 11 CNR comparison, edge-preserving prior (TV) vs non-ideal anatomical priors, calculated on 25 noise realizations.
Fig. 10 CRC and CNR comparison, edge-preserving prior (TV) vs ideal anatomical priors, calculated on 25 noise realizations.
For a fair comparison of edge and node clustering, we use edge-GDV-similarity and node-GDV-similarity, conceptually equivalent measures of the network position of an edge and a node, respectively, as distance metrics for the same clustering method.
This approach allows quantitative comparison of edge importance, and sheds light on the parametric dependence of relative edge importance, for instance in a voltage-gated ion channel.
Fig. 9 Comparison of edge detectors: a recognition accuracy on UCF Sports, b recognition accuracy on YouTube, and c Example frames of different edge detectors.
They're showing a comparison with edge.
For the skeleton comparison, an edge is counted as a true positive as long as it has the correct link.
In other words, significant proteins were identified by the comparison between edge variations of the PPI networks in the two stages of the infection process, as revealed by the SVVs in the differential PPI networks with P values ≤ 0.05 (distributions of SVVs for C. albicans and zebrafish were shown in Supplementary Figure S3).
Modularity function measures the density of edges inside communities in comparison to edges between communities [49].
Initially, for a fair comparison with edge-SN, we cluster adjacent edges only, to test if and how much we gain by using our more sensitive measure of edge similarity.
Mann-Whitney pairwise comparisons reveal that nearly all pairwise comparisons between edge number groups as suggested by box-plot notches in Figure 6 (P <0.05) (see Additional file 1 for pairwise comparisons).
