### Table 1. Sample programs used in this paper. Sizes and basic block counts are reported for unoptimized and size-optimized compilation, respectively

2003

Cited by 4

### Table 1: Optimal number of samples s for sorting the [WR] integer benchmark on the Cray T3D, for a variety of processors and input sizes.

1996

"... In PAGE 27: ...307 1.61 Table1 0: Total execution time (in seconds) to sort the [U] integer benchmark on the IBM SP-2, com- paring our results (HJB) with those reported by Gerbessiotis and Siniolakis (GS) . our regular sampling algorithm, none of the steps in our sample sort algorithm exhibit such strong dependence on p.... In PAGE 28: ...3 % 5.0 % Table1 1: Comparison of time required by our regular sampling algorithm with the time required by our sample sort algorithm using our [WR] benchmark. If T RS#28n; p#29 represents the time required by our regular sampling algorithm and T SS#28n; p#29 represents the time required by our random sample sort algorithm, then the corresponding entry is #28T RS#28n; p#29 , T SS#28n; p#29#29 as a percent- age of T SS#28n; p#29.... ..."

Cited by 20

### Table II: Optimal number of samples s for sorting the [WR] integer benchmark on the IBM SP-2-WN, for a variety of processors and input sizes.

1996

Cited by 20

### Table 2: Optimal number of samples s for sorting the [WR] integer benchmark on the IBM SP-2-WN, for a variety of processors and input sizes.

1996

Cited by 20

### Table I: Optimal number of samples s for sorting the [WR] integer benchmark on the Cray T3D, for a variety of processors and input sizes.

1996

Cited by 20

### Table II: Optimal number of samples s for sorting the [WR] integer benchmark on the IBM SP-2-WN, for a variety of processors and input sizes.

1996

Cited by 20

### Table 3. Subsampling methods: trade-off among the values of k, the number of partitions B, and the sample size, S. Last column denote the percentage of sample size regarding the entire data set. (Bold represents most optimal)

2004

"... In PAGE 6: ... This implies that in a large data set, a small fraction of data can be representative of the entire data set, a result that holds great computational promise for distributed data mining. The optimal sample size, S, and granularity of the component partitions derived by subsampling are reported in Table3 . We see that the accuracy of the resampling method is very similar to that of the bootstrap algorithm, as reported in Table 2.... In PAGE 6: ... We see that the accuracy of the resampling method is very similar to that of the bootstrap algorithm, as reported in Table 2. This equivalent level of accuracy was reached with remarkably smaller sample sizes and much lower computational complexity! The trade-off between the accuracy of the overall clustering combination and computational effort for generating component partitions is shown in Table3 , where we compare accuracy of consensus partitions. The most promising result is that only a small fraction of data (i.... ..."

Cited by 6

### Table 2: Description of database properties and sample error results when optimizing the pruned tree. Attributes Database Size Error Cont 1 2 3 4 5 6 gt; 6 Miss Tot

1995

"... In PAGE 5: ... Robert Detrano. Table2 lists the properties of the individual databases. The number in parentheses next to the Database label is the number of di erent class values appearing in the data.... In PAGE 6: ...number of splits used for each tree was also recorded, but only the error was optimized. The last three columns of Table2 show a sample of the results when optimizing the pruned tree produced by C4.5 according to the hill- climbing optimization strategy described in Section 4).... ..."

Cited by 7

### Table 4: Averages and standard deviations of the numbers of configurations stored with a roadmap edge after the optimization of roadmap path segments. The data is the combination of size constrained and time constrained runs. Sample size is 30 runs.

"... In PAGE 45: ...0001) for both sets. Table4 presents descriptive statistics of the number of configurations stored with the roadmap edges after a path segment produced with a powerful local planner has been optimized. The data shows that on the average only a small number of configurations have to be stored in each edge.... ..."

### Table 1. Results of the optimizations

1998

"... In PAGE 12: ...Table 1. Results of the optimizations Table1 shows the e ect of using the above optimizers on the size of our sample application and the time to generate the FSMs 4. Each line adds one optimizer to the FSM generation and gives the corresponding results.... ..."

Cited by 13