Information Technology Reference
In-Depth Information
Table 1. Clusters used in the large experiment
e 50 10
f 45 5 10
g 405510
h 3555510
i 30555510
j 30 10 5 5 10
k 255555510
l 251055510
m 25 10 10 5 10
n 2055555510
o 2010555510
p 20 10 10 5 5 10
q 20 10 10 10 10
Fig. 4. Difference of entropies, experiments e through i , o through q
was 50% of the entire network, to which a subsequent cluster of size equal to
10% of the network was then added. In the last experiment, we began with a
single cluster that was 20% of the network and then added three clusters each
of size equal to 10% of the network before adding a final cluster of size 10%.
We apologize for less-than-optimal plotting capability in this version of the
paper, but present the results below. Figures 4 and 5 contain the differences of
mutual entropy of Shannon and Renyi, and is thus a generalization of Figure
3. Although the sequences are somewhat hard to distinguish, the lines repre-
senting entropy values are in essentially reverse order, top to bottom, as the
cluster sequences are presented in Table 1. The upper grouping in Figure 4 is
of experiments i through e , top to bottom, and the lower grouping is of experi-
Search WWH ::




Custom Search