This implies that relaxed graph contrast results, for larger sets, yield lower values than relaxed set contrast results.
The graph contrast uses subtrees and the set contrast uses sets associated with these subtrees, as there is a one-to-one relationship between subtree and its set representation, the number of elements to contrast in both methods is the same.
Cases of these are for example situations where it is known that hundreds of substances are potential candidates for biological screening for antibreast cancer, in this case the graph contrast would look for the different graph structures for those substances, which would indicate which substances can be sent first for screening, namely those which appear in different subtrees of the cluster.
Unlike graph and relaxed-graph contrasts, often set and relaxed-set contrast frequencies attained a value of 1, the latter case being the most striking one.
It was observed that frequencies calculated through set and graph cluster contrasts spread clusters and their frequencies along particular frequency values, which is depicted as series of dots along horizontal lines (Fig. 8).
For a spatial distribution of these graph theoretical contrasts (hits vs. misses, normalized change), see Figures 4 A and 5. Node degree and betweenness centrality show a strong spatial overlap regarding the left auditory cortex (Figs 4 A and 5).
The relaxed graph cluster contrast uses more elements for the contrast than the relaxed set cluster contrast.
Any conclusion can be drawn from the finding that as relaxed graph the contrast of a cluster in a dendrogram is 0.03 while as a relaxed set the contrast is 0.3.
The relaxed graph cluster-contrast function, unlike graph-cluster one, produces frequencies greater than 0.5 for large clusters.
