### Table 1: Summary of the characteristics of the benchmarks. The stock management problems theoretically lead to convex Bellman-functions, but their learnt counterparts are not convex. The arm and away problem deal with robot-hand-control; these two problems can be handled approximately (but not exactly) by bang-bang solutions. Walls and Multi-Agent problems are motion-control problems with hard penalties when hitting boundaries; the loss functions are very unsmooth.

### Table 1: Unsmoothed and Smoothed Confusabilities.

2000

"... In PAGE 6: ... Consider for instance the word Boston. Table1 shows the five most con- fusable lexemes of each word, computed from (26) both without... ..."

Cited by 6

### TABLE I. Unsmoothed and smoothed confu-

2000

Cited by 6

### Table 4.9. Performance of the unsmoothed model on various syntactic constructions

2004

Cited by 1

### Table 2: Simulation Results for unsmoothed Star Wars trace without and with Importance Sampling.

1997

"... In PAGE 13: ... This fact has to be accounted for when computing the sample variance (see Figure 2, line 12) since we have only L independent trials. In Table2 (column no IS) we give the sample mean and 90% con dence intervals for Ptime loss for the unsmoothed Star Wars trace. In this experiment we ran L = 500 replications for a total of 500 174; 136 87 106 simulated frame times.... In PAGE 14: ... reduction techniques, such as Importance Sampling (discussed below), would require immense computational resources. Each of the 4 simulation results given in Table2 took already about 2 days on a SPARCstation 2. Table 3 shows the results obtained after smoothing the Star Wars trace over each Group of Pictures (GOP).... In PAGE 15: ... This distribution encourages large values for the xm apos;s and leads to more frequent losses; the g( ) in the denominator of the expression for h( ) compensates for this e ect. The IS results displayed in Table2 were obtained for L = 2 106; note that the IS has signi cantly reduced the width of the con dence intervals. Moreover, each of the four simulation results took approximately 2 hours on a SPARCstation 2.... In PAGE 20: ... Figure 6 shows the number of admissible Star Wars connections computed with varying resolutions (binsizes) for the Star Wars histogram. Table2 gives the typical CPU times required for an admission test. All computations are performed on a Sun SPARCstation 2.... ..."

Cited by 18

### Table C2. Unsmoothed Data for Figure 5, Azimuth Scans at 20 Elevation

1997

Cited by 1

### Table C3. Unsmoothed Data for Figure 6, Elevation Scans at 141 Azimuth

1997

Cited by 1

### Table 1 Square Root of the Average Squared Difference Between Smoothed and Unsmoothed Standard Errors

"... In PAGE 3: ... Standard error estimates from VPLX tend towards the middle of the plot. Table1 shows the mean of the stability measure for each publishable index series. For the All Items category, SUDAAN has the most stable estimates.... ..."

### Table 1. Results of an ABX type evaluation comparing the smoothed and unsmoothed synthetic speech.

### Table 1. Shape comparison results for the maximum t-statistic permutation test for the unsmoothed and smoothed dataset. Volume and area results for comparison.

2007

Cited by 4