### Table 4: 3{dimensional distributions de ning the decomposable model. Just as every Bayesian network can be converted into a decomposable model, a decomposable model can always be expressed in the form of a Bayesian net- work. To do so, the nodes of the decomposable model (i.e. the indices of the variables) are ordered so that the corresponding ordering of the cliques meets the running intersection property: 1; : : :; k | {z } C1

1997

Cited by 2

### Table 2: E ect of clique cuts

1998

"... In PAGE 16: ... In order to understand how much of the performance can be attributed to clique cuts, we ran MINTO on the same set of instances with clique generation turned o . In Table2 , we present the results. This table shows that for almost all of the instances, performance degrades drastically without the clique cuts.... ..."

Cited by 7

### Table 2. Gradient descent clique weight vector estimates using a 1st order, GGMRF MPE estimator for various input MRF realizations.

"... In PAGE 7: ... The images of Figure 8 yield coe cient estimates with q = 1 shown in Table 1 which are quite di erent in some cases from those which are typically assigned ad hoc. In Table2 we present the results of coe cient estimates from Monte Carlo simulations of images with GGMRF distributions and q = 1. The optimization is performed via gradient descent of the risk function, following initializa- tion by the \quantized quot; coe cient estimate described above for the estimator under q = 1.... ..."

### Table 1: Average total clique size. number of solved tasks

"... In PAGE 7: ... Table1 and Figure 5 asummarize the results. Pro- vided numbers are the average of the total clique size for all possible orderings of tasks.... ..."

### Table A.1 Maximum clique sizes (C) and maximum separator widths (S) for the join trees obtained by tree clustering with various ordering heuristics Circuit #variables Causal Maximum Minimum Minimum

1996

Cited by 45

### Table A.1 Maximum clique sizes (C) and maximum separator widths (S) for the join trees obtained by tree clustering with various ordering heuristics Circuit #variables Causal Maximum Minimum Minimum

1996

Cited by 45

### Table 4. Average rank for various clique sizes from Table 1

1999

"... In PAGE 2: ... As the aver- age rank increases with neighbourhood size, we can sur- mise that a small neighbourhood is better for classification. In Table4 it is the statistical order that is varied. From this table we can see that although it is advisable to keep the statistical order small, if the the statistical order gets too small the model will start to be undertrained.... ..."

Cited by 10

### Table 4. Average rank for various clique sizes from Table 1

1999

"... In PAGE 2: ... As the aver- age rank increases with neighbourhood size, we can sur- mise that a small neighbourhood is better for classification. In Table4 it is the statistical order that is varied. From this table we can see that although it is advisable to keep the statistical order small, if the the statistical order gets too small the model will start to be undertrained.... ..."

Cited by 10

### Table 4-1 Summary of selected SNA metrics. Ego-betweenness Closeness Maximal Clique Algorithm

2007

"... In PAGE 9: ...able 1-1. The social data about friendships between students in a small class. .4 Table4 -1.... In PAGE 68: ...ifficulty. The time cost for this metric is about the third order of graph size. We chose maximal clique enumeration problem as our test case for the hardest problems because in the worst case, finding all maximal cliques will take exponential costs on both time and memory. Properties of selected SNA metrics and their corresponding algorithms are summarized in Table4 -1. ... ..."

### Table 7: Clique Partitioning and Memory- Binding Comparison for the DCT Benchmark

"... In PAGE 6: ... In addition, the number of multiplexers is reduced since memory binding assigns more variables to the same memory; thus reducing the number of memories used. In order to illustrate the difference between memory binding and the other register binding techniques, Table 6 and Table7 show a comparison between the results for Memory ... ..."