### Table 5: Results for the minimum cut problem in ToR graphs. Type Date Opt Time #BN % 1 % gt;10

"... In PAGE 25: ... The results of these computations can be found in Table 5. Table5 gives the results for the ToR graphs. The rst two columns show the graph type and the date, the third column contains the optimal value of the minimum cuts, and nally we give the computation times (in seconds),... In PAGE 26: ... Again, all values are average values over all 1081 pairs of ASs for a speci c graph type and date. As can be seen from Table5 , the algorithm for nding the minimum cut sizes in ToR graphs is very fast, also compared to the computation times for the algorithms for nding the maximum number of vertex-disjoint paths. Again, the number of branching nodes needed to nd the integer optimum is small, since the solution to the LP-relaxation is integral in 98.... ..."

### Table 4: Clique-size performance and running time of MFA-CM on N-vertex p-random graphs, averaged over R graphs in each row. N=R p

1996

"... In PAGE 7: ... We chose 100-vertex 0.9-random graphs both so as to evaluate many more parameter settings and also to test if the best parameter settings would later scale up to larger graphs (see Table4 ). Table 3 reports our experiments.... In PAGE 8: ... We then tested if the parameter setting P1(N) would scale up to larger graphs and the running time of MFA-CM with this parameter setting on these graphs. Table4 reports the results. The rows with the same values of N and p were tested on identical graphs.... In PAGE 8: ... (In earlier experiments [JR92] on the same ten 400-vertex 0.9-random graphs as in Table4 , we obtained an average clique size of 50.4 with a slow geometric annealing schedule and an average clique size of 43.... ..."

Cited by 15

### Table 1: Comparison of edge classi cations. ToR Graphs Percentages of identically classi ed edges 18.04.2001 04.02.2002 06.04.2002 09.01.2003 10.02.2004

"... In PAGE 21: ... Thus, it is interesting to see how many edges between ASs are classi ed in the same way by the di erent algorithms. In Table1 the percentages of identically classi ed edges are given for all six combinations of ToR graphs. For example, 95.... In PAGE 34: ...Table1 0: Percentage of pairs of ASs connected by customer chains. Date A B C D 18.... ..."

### Table 10: Percentage of pairs of ASs connected by customer chains. Date A B C D

"... In PAGE 34: ...given in Table10 . This table shows for each of the ve dates the percentages of pairs of ASs that are connected by customer chains in all our types of ToR graphs.... ..."

### Table 1. Span and depth of complete objectsa

2002

"... In PAGE 2: ...8.71 mm, was calculated from 319 sections * 0.09 mm section21.) The average number of sections that a cone pedicle or cell soma spans was measured for all complete individuals of each type in the series of 319 sections ( Table1 ). For a pedicle that was only partially within the series, we identified its center section as its end section within the series plus or minus half of the average span of a pedicle, 80.... In PAGE 4: ...group of cell ( Table1 ). For GCL cells, we obtained coordinates for the center of their nucleolus or the center of the nucleus in the few cases in which there were two nucleoli.... ..."

### Table 12: Relative error reductions (ERREL) for each collective inference method across all data sets and relational classifier methods. (The three collective inference methods are relaxation labeling (RL), iterative classification (IC), and Gibbs sampling (GS)). The last column, overall, is the average error reduction for each method across all sample ratios. Bold entries indicate the largest relative error reduction for each sample ratio.

"... In PAGE 28: ...where err(RC-CI,D, r) is the error for the RC-CI configuration (a relational classifier paired with a collective inference method) on data set D with r% of the graph being labeled. Table12 shows the relative error reductions for each collective inference component. Relaxation labeling outperformed iterative classification across the board, from as low as a 1.... ..."

### Table 7: Vertex-disjoint paths in ToR and undirected graphs. Graph Type avg VDP min VDP max VDP

"... In PAGE 28: ...4.1 Connectivity measures for the Internet In Table7 we compare the number of vertex-disjoint paths for the di erent types of ToR graphs, the undirected BGP graphs and the CAIDA graph. In the second column we give the average number of vertex-disjoint paths, averaged over all pairs of ASs and all dates of the speci ed graph type.... ..."

### Table C9 Proportion of the population in each birth cohort with secondary complete

1999

Cited by 2

### Table 5: A Truth Table to Example 13.2 Proceeding in the same way for the output-consistency relation CON, we obtain the following r- partitions:

1995

"... In PAGE 73: ...f F using an intermediate variable g. The truth table obtained in this way is shown in Table 7. Example 15.2 For another example, in Table5 , we have: P(B) = P(X1X2) = ff0; 6; 7g; f1; 6; 7g; f2; 3; 5; 7g; f2; 3; 4; 7gg; P(A) = P(X0) = ff0; 1; 2; 4;5g; f2; 3; 4; 5;6; 7gg; if G = ff0; 1; 6; 7g; f2; 3; 4; 5; 7gg;... In PAGE 74: ... Example 15.3 For example, in Table5 , repeated in Table 8, we have: P(X1 X2) = P(B) = ff0; 6; 7g; f1; 6; 7g; f2; 3; 5; 7g; f2; 3; 4; 7gg = fB1; B2; B3; B4g; If we merge B1 and B2 together, we get: P12(B) = ff0; 1; 6; 7g; f2; 3; 5; 7g; f2; 3; 4; 7gg; P(A) P12(B) = ff0; 1g; f6; 7g; f2; 3; 5; 7g; f2; 3; 4; 7gg lt; PF therefore B1 and B2 are compatible, denoted as B1 B2. The same way we can check the compatibility relation of other blocks in P(B).... ..."

### Table C1 Proportion of the population in each birth cohort completing at least some primary

1999

Cited by 2