### Table 2: Con dence Intervals for the Parameters of the Normal-Normal Model for the

1999

"... In PAGE 7: ... Comparing the tted curves in these plots to the actual data suggests that the normal-normal and normal-sev models t the data well. Table2 gives approximate con dence intervals for the normal-normal model parameters based on large-sample asymptotics and likelihood ratio methods for the laminate panel data. It also gives the asymptotic standard errors and coe cients of variation (the standard error as a percentage of the estimate) of the estimators.... ..."

Cited by 1

### Table A3: Factor Loadings of Rotated Component Matrix* No Variables Factors 1 2 3

2003

### Table 4 Normalized weights and weight intervals for the second level of the objectives hierarchy

in based

2002

"... In PAGE 9: ... Note that, again, it is possible to provide minimum and maximum values to get imprecise assessment by means of weight intervals. Table4 and Fig. 9 show the weight intervals and the (precise) normalized average intervals for the attributes and objectives, which are automatically computed by the system by means of the formulas (see Refs.... ..."

### Table 4. Ccmpari son of Adjustment Methods Using Both Weighted

"... In PAGE 23: ... Numbers in parentheses are counts of states falling in the interval. Models (5a) and (6a) in Table4 consist of predictions based on equations 5 and 6. Models (5b) and (6b) are mixtures of the direct estimates and their respective modelled predictions.... In PAGE 23: ... Models (5b) and (6b) are mixtures of the direct estimates and their respective modelled predictions. According to Table4 model (6b) appears to be the best although not by very much. Since model (5b) using the minority renter variable does almost as well as the three variable model (both using estimated standard errors) it probably could be used without much if any loss in the precision of the adjustment results.... ..."

### Table 2: Con dence Intervals for the Parameters of the Normal-Normal Model for the Laminate Panel Data

1999

"... In PAGE 7: ... Comparing the tted curves in these plots to the actual data suggests that the normal-normal and normal-sev models t the data well. Table2 gives approximate con dence intervals for the normal-normal model parameters based on large-sample asymptotics and likelihood ratio methods for the laminate panel data. It also gives the asymptotic standard errors and coe cients of variation (the standard error as a percentage of the estimate) of the estimators.... ..."

Cited by 1

### Table 6: Con dence Intervals for the Parameters of the Normal-Normal Model for the Nickel-Base Superalloy Data

1999

"... In PAGE 21: ... Although not as obvious, a similar observation can be made for the laminate panel example in Section 4. Table6 gives approximate con dence intervals for normal-normal model parameters based on large-sample asymptotics and likelihood ratio methods. It also gives the asymp- totic standard errors and coe cient of variation of the estimators.... ..."

Cited by 1

### TABLE 4. Producing Ideas: At+1 = t L t A

1999

Cited by 10

### TABLE 1 MODEL SIZE AND RELATIVE ELAPSED TIMES AS A FUNCTION

1993

Cited by 1

### Table 6: Comparison of the best overall solution found by each one of the methods included in MOSES for the rst example (design of a 25-bar space truss). GA-based methods were tried with binary (B) and oating point (FP) representations. The following abbreviations were used: OS = Osyczka apos;s System, GCM = Global Criterion Method (exponent=2.0), WMM (Weighting Min-max), PWM (Pure Weighting Method), NWM (Normalized Weighting Method), GALC = Genetic Algorithm with a linear combination of objectives using scaling. In all cases, weights were assumed equal to 0.33 (equal weight for every objective).

2000

"... In PAGE 19: ...i.e., the total weight of the truss) in a discrete manner. Assuming continuous variables, the GA-engine for single objective optimization was able to nd a lighter truss. As we can see in Table6 , the new GA-based approach proposed by the authors, named GAminmax, provide the best overall results when a oating representation was used. It should be noted that our approach performs hardly over the average when binary representation is used.... ..."

Cited by 11

### Table 2: BLEU scores and brevity penalty (BP) factors determined on the ZHEN-06 test set for primary systems together with consensus systems for the MBR-like candidate selection method obtained by combining each three adjacent systems with uniform system prior weights. Primary systems are sorted in descending order with respect to their BLEU score. The 95% confidence intervals are computed using the bootstrap re-sampling normal approximation method (Noreen, 1989).

2007

"... In PAGE 5: ... Each translation system corresponds to a primary system submitted to the NIST MT06 evaluation 3. Table2 shows the BLEU scores together with their corresponding BP factors for the primary systems of 16 research labs (site names were anonymized). Primary systems are sorted in descending order with respect to their BLEU score.... In PAGE 5: ... Primary systems are sorted in descending order with respect to their BLEU score. Table2 also shows the consensus translation results for the MBR-like candidate selec- tion method. Except where marked with an asterisk, all consensus systems are built from the outputs of three adjacent systems.... In PAGE 5: ... Improvements are considered to be significant if the left boundary of the confidence interval is larger than zero. Oracle BLEU scores shown in Table2 are com- puted by selecting the best translation among the three candidates. The oracle scores might indicate a larger potential of the MBR-like selection rule, and further gains could be expected if the candidate se- lection rule is combined with confidence measures.... In PAGE 5: ... The oracle scores might indicate a larger potential of the MBR-like selection rule, and further gains could be expected if the candidate se- lection rule is combined with confidence measures. Table2 shows that it is important that all trans- lation systems achieve nearly equal quality; com- bining high-performing systems with low-quality translations typically results in clear performance losses compared to the primary system, which is the case when combining, e.... ..."

Cited by 1