### Table 1: Performance of multi-stage clustering on the male subset of the 1998 Switchboard development test set.

2000

"... In PAGE 3: ...obtained in the two systems were similar, as were the recognition results (see Table1 ). The results show that incorporating features in multiple stages is a viable method for using a large number of... In PAGE 3: ... The syllable and word information could be asked about the center and neighboring phones. Note that the baseline triphone and pentaphone systems shown here used a larger search space than in Table1 (C5 BP BGBC vs. C5 BP BDBH), hence the slightly better performance.... ..."

Cited by 10

### Table 2: Results obtained by the multi-stage GFRBSs in the problem being solved Generation Multisimpli cation Tuning

1999

"... In PAGE 23: ... The SE values over these two sets are labeled as training and test. In this case, we de ne SE as 1 2 N N Xi=1(~ li ? li)2 The results obtained in the di erent experiments performed with the GFRBS considered are collected in Table2 where #R stands for the number of rules in the corresponding FRB, and SEtra and SEtst for the values obtained in the SE measure computed over the training and test data sets, respectively. In view of the results obtained, we should remark on: a) the good performance of the genetic multisimpli cation process since, in the second and third iterations, it allows us to generate fuzzy models with better approximate and predictive behavior, i.... ..."

Cited by 18

### Table 1: Performance of multi-stage clustering on the male subset of the 1998 Switchboard development test set.

"... In PAGE 3: ...obtained in the two systems were similar, as were the recognition results (see Table1 ). The results show that incorporating features in multiple stages is a viable method for using a large number of... ..."

### Table 1: Performance of multi-stage clustering on the male subset of the 1998 Switchboard development test set.

"... In PAGE 3: ...obtained in the two systems were similar, as were the recognition results (see Table1 ). The results show that incorporating features in multiple stages is a viable method for using a large number of... In PAGE 3: ... The syllable and word information could be asked about the center and neighboring phones. Note that the baseline triphone and pentaphone systems shown here used a larger search space than in Table1 (M = 40 vs. M = 15), hence the slightly better performance.... ..."

### Table 4 Results of the GHH, multi-stage GHH and the other 3 approaches in literature on benchmark course

2007

"... In PAGE 18: ... We test our approach with the same number of runs as that of the Hyper-heuristic from (Burke, Kendall and Soubeiga, 2003) to make a more fair comparison. The term x% Inf in Table4 indicates the percentage of runs which failed to obtain feasible solutions. From Table 4 we can observe that our GHH approach obtains competitive results with the other 3 approaches on these course timetabling problems.... In PAGE 18: ... The term x% Inf in Table 4 indicates the percentage of runs which failed to obtain feasible solutions. From Table4 we can observe that our GHH approach obtains competitive results with the other 3 approaches on these course timetabling problems. For problem Medium5 , it obtained the best results among all approaches.... ..."

Cited by 8

### Table 4 Results of the GHH, multi-stage GHH and the other 3 approaches in literature on benchmark course timetabling problems

2007

"... In PAGE 12: ... We test our approach with the same number of runs as that of the Hyper-heuristic from (Burke, Kendall and Soubeiga, 2003) to make a more fair comparison. The term x% Inf in Table4 indicates the percentage of runs which failed to obtain feasible solutions. From Table 4 we can observe that our GHH approach obtains competitive results with the other 3 approaches on these course timetabling problems.... In PAGE 12: ... The term x% Inf in Table 4 indicates the percentage of runs which failed to obtain feasible solutions. From Table4 we can observe that our GHH approach obtains competitive results with the other 3 approaches on these course timetabling problems. For problem Medium5 , it obtained the best results among all approaches.... ..."

Cited by 8

### Table 4 Results of the GHH, multi-stage GHH and the other 3 approaches in literature on benchmark course timetabling problems

2007

"... In PAGE 11: ... We test our approach with the same number of runs as that of the Hyper-heuristic from (Burke, Kendall and Soubeiga, 2003) to make a more fair comparison. The term x% Inf in Table4 indicates the percentage of runs which failed to obtain feasible solutions. From Table 4 we can observe that our GHH approach obtains competitive results with the other 3 approaches on these course timetabling problems.... In PAGE 11: ... The term x% Inf in Table 4 indicates the percentage of runs which failed to obtain feasible solutions. From Table4 we can observe that our GHH approach obtains competitive results with the other 3 approaches on these course timetabling problems. For problem Medium5 , it obtained the best results among all approaches.... ..."

Cited by 8

### Table 2: Results obtained by the multi-stage GFRBSs in the electrical rural problem

2000

"... In PAGE 16: ... Genetic tuning process: N = 61, Pc = 0:6, Pm = 0:1, a = 0:35, b = 5, d = 0:001, 1000 generations, 25 (1+1)-ES iterations, = 0 and c = 0:9 (the updating amount of the Rechenberg apos;s 1=5-success rule in the (1+1)-ES [3]). Tables 2 and 3 collect the results obtained by the proposed method comparing with several MOGUL methods and other learning approaches, respectively: In Table2 , we can see as the proposed method gets the best results for this complex problem containing noise. It can be stand out that the best result is obtained for M!TSK after the tuning stage, with a 17:3% of improving in test and a 12:7% in training over the best result of the remaining MOGUL methods.... ..."

Cited by 1

### Table 4 Results of the GHH, multi-stage GHH and the other 3 approaches in literature on benchmark course

"... In PAGE 18: ... We test our approach with the same number of runs as that of the Hyper-heuristic from (Burke, Kendall and Soubeiga, 2003) to make a more fair comparison. The term x% Inf in Table4 indicates the percentage of runs which failed to obtain feasible solutions. From Table 4 we can observe that our GHH approach obtains competitive results with the other 3 approaches on these course timetabling problems.... In PAGE 18: ... The term x% Inf in Table 4 indicates the percentage of runs which failed to obtain feasible solutions. From Table4 we can observe that our GHH approach obtains competitive results with the other 3 approaches on these course timetabling problems. For problem Medium5 , it obtained the best results among all approaches.... ..."