### Table 1: Performance of the canonical structure prediction compared to the usual predic- tion of RNAfold. For the mfes, the values compared where the MCCs, for the partition functions, see text for comparison strategy. TP = true positive, FP = false positive. no LP = canonical structures. .

### Table 2: Percent of words pronounced canonically for phonetic and hybrid lexical representations

"... In PAGE 4: ... Often there is enough residual phonetic evidence of the deleted phone, or enough phonetic evidence in other parts of the word, to rec- ognize a word correctly despite the deletion. Thus, we decided to use a two-part strategy in calculating canonical pronunciation ( Table2 ). The first column, strict matching , allows no insertions or deletions when comparing the canonical and realized pronun- ciation.... In PAGE 4: ...) 4.2 Results and Discussion In Table2 , we see that a standard ASR lexicon ap- proach (strict matching 1), does not match the tran-... ..."

### Table 5: Average time to compute the representative of a state.

"... In PAGE 7: ... We apply three symmetry reduction strategies: full is a canonical strategy which computes every possible permuta- tion of a state, segmented is canonical strategy which uses a heuristic for a more efficient representative computation, and sorted is a normalizing strategy. Table5 depicts a comparison of the time required for computing the representative of a state, which shows that sorted is the fastest one. Of course, the price to pay is a weaker space reduction.... ..."

### Table 5: Average time to compute the representative of a state.

"... In PAGE 7: ... We apply three symmetry reduction strategies: full is a canonical strategy which computes every possible permuta- tion of a state, segmented is canonical strategy which uses a heuristic for a more efficient representative computation, and sorted is a normalizing strategy. Table5 depicts a comparison of the time required for computing the representative of a state, which shows that sorted is the fastest one. Of course, the price to pay is a weaker space reduction.... ..."

### Table 1: Results for Hard Problems 6 Conclusions and future directions In this paper, we have discussed an e cient im- plementation strategy to obtain the canonical cover of a logic function. The notion of formulation of the set covering problem without explicitly generating the primes in the set is attractive since it reduces the size of the covering table signi cantly. The full import of the essential signature cube tech- niques will be realized when we move towards the mul- tilevel logic synthesis which has very large don apos;t care set at each node of the boolean network. References

### Table 1: Statistical characteristics of datasets: the number of examples (Ex. No.), the resampling strategy (Strategy) (tt = train and test, 9x = 9 fold cross-validation), the number of attributes (Att. No.), the number of binary/categorical attributes (Cat. Att.), the number of classes (Classes), the homogeneity of covarience (SD ratio), mean absolute correlation coe cient ( ), cannonical discriminant correlation (cancor), variation explained by rst four canonical discriminants (fract), average skew (skew), and average kurtosis (kurtosis).

"... In PAGE 20: ... 5.8 Diabetes data This dataset has no categorical attributes ( Table1 0). The three discriminants perform well, and so does SMART.... In PAGE 20: ... 5.9 Heart disease and head injury Both Head ( Table1 2) and Heart (Table 11) have cost matrices associated with them. In the results for both datasets the top 10 algorithms (algorithms with the lowest costs) are all capable of utilizing costs in the testing phase.... In PAGE 21: ... 5.10 German credit data This data ( Table1 3) has similar characteristics to those of the other credit dataset, but di ers in having a cost matrix (so lower cost is desired), and higher skew (1.6986) and kurtosis (7.... In PAGE 23: ...1 Symbolic learning Symbolic learning methods performed best (in terms of accuracy) on the Shuttle, Segment, and Credit datasets. Examining Table1 , we note that Shuttle and Segment have the highest... In PAGE 29: ... This is true for these two medical datasets. From Table1 the correlation between attributes in Head is ( = 0:1217), Heart ( = 0:1236), and Diabetes ( = 0:1439). All are less than 0:15.... In PAGE 67: ...10 65.10 { { Table1 0: Diabetes data results (Alloc80, marked by \@ quot;, achieved its best results using only attributes 2, 8, 6, and 7. Cal5, marked by \# quot;, used the newer version of Cal5.... In PAGE 68: ...844 1.78 12* NA Table1 1: Heart disease results... In PAGE 69: ...0 213.5 49 1 Table1 2: Head injury results... In PAGE 70: ...340 2.16 182 12 Table1 3: German credit data results (Alloc80, marked by \@ quot;, achieved its best results using attributes 1, 3, and 13. CART apos;s times, marked by \# quot;, was quoted for a Solbourne sparc clone)... ..."

### Table I. Comparison of Query Languages Concerning Tie

### Table 6. Canonical redundancy analysis Canonical

2003

"... In PAGE 7: ... Although the first and second canonical functions are ignificant according to the above analysis, it is mmended that redundancy analysis be utilized to etermine which functions should be used in the terpretation [37]. Redundancy is defined as the ability of f independent variables, taken as a set, to explain the ariation in the dependent variables taken one at a time Table6 summarizes the redundancy analysis for the dent and independent variables for the two canonical unctions that were found to be significant by using the easure of model fit. The results indicate that the first onical function accounts for the highest proportion of otal redundancy (93.... ..."

Cited by 2

### Table 6. Canonical redundancy analysis Canonical

2003

"... In PAGE 7: ... Although the first and second canonical functions are ignificant according to the above analysis, it is mmended that redundancy analysis be utilized to etermine which functions should be used in the terpretation [37]. Redundancy is defined as the ability of f independent variables, taken as a set, to explain the ariation in the dependent variables taken one at a time Table6 summarizes the redundancy analysis for the dent and independent variables for the two canonical unctions that were found to be significant by using the easure of model fit. The results indicate that the first onical function accounts for the highest proportion of otal redundancy (93.... ..."

Cited by 2