Skip to Main Content
Table 11:
Analysis of algorithm power through the predicted accuracy, ACC, area, , density, , and purity, , of the solution and absolute footprints as predicted by the models validated in Section 4, and their difference with the results based on experimental data. The area and density are normalized using the values from the known region of the instance space as references, i.e., and . In boldface are the differences that exceed 10%.
Solution footprintAbsolute footprint
AlgorithmACCACC
 BFGS 90.6% 44.1% 91.4% 99.5% 77.3% 2.7% 514.7% 99.0% 
 BIPOP-CMA-ES 97.3% 97.4% 98.3% 99.4% 92.2% 23.8% 176.0% 98.7% 
 LSstep 87.7% 3.5% 511.6% 99.6% 66.7% 3.5% 241.2% 100.0% 
Predicted Nelder–Doerr 91.9% 62.8% 85.8% 99.3% 86.7% 36.7% 67.8% 97.2% 
 CMA-ES 82.1% 60.1% 97.6% 99.8% 77.1% 0.8% 787.4% 100.0% 
  
 -easy 96.1% 62.8% 85.8% 99.3% 
 -hard 85.6% 22.5% 177.9% 94.6% 
 BFGS  3.3% 1.0% 0.6%  0.5% 109.6% 0.6% 
 BIPOP-CMA-ES  0.9% 0.8% 0.1%  4.8% 16.3% 1.5% 
 LSstep  7.1% 321.5% 1.3%  0.5% 47.5% 8.1% 
Difference Nelder–Doerr  2.4% 0.6% 0.2%  5.1% 5.6% 6.9% 
 CMA-ES  3.2% 0.6% 0.4%  0.3% 244.6% 2.4% 
  
 -easy  2.4% 0.6% 0.2% 
 -hard  1.7% 5.1% 0.1% 
Solution footprintAbsolute footprint
AlgorithmACCACC
 BFGS 90.6% 44.1% 91.4% 99.5% 77.3% 2.7% 514.7% 99.0% 
 BIPOP-CMA-ES 97.3% 97.4% 98.3% 99.4% 92.2% 23.8% 176.0% 98.7% 
 LSstep 87.7% 3.5% 511.6% 99.6% 66.7% 3.5% 241.2% 100.0% 
Predicted Nelder–Doerr 91.9% 62.8% 85.8% 99.3% 86.7% 36.7% 67.8% 97.2% 
 CMA-ES 82.1% 60.1% 97.6% 99.8% 77.1% 0.8% 787.4% 100.0% 
  
 -easy 96.1% 62.8% 85.8% 99.3% 
 -hard 85.6% 22.5% 177.9% 94.6% 
 BFGS  3.3% 1.0% 0.6%  0.5% 109.6% 0.6% 
 BIPOP-CMA-ES  0.9% 0.8% 0.1%  4.8% 16.3% 1.5% 
 LSstep  7.1% 321.5% 1.3%  0.5% 47.5% 8.1% 
Difference Nelder–Doerr  2.4% 0.6% 0.2%  5.1% 5.6% 6.9% 
 CMA-ES  3.2% 0.6% 0.4%  0.3% 244.6% 2.4% 
  
 -easy  2.4% 0.6% 0.2% 
 -hard  1.7% 5.1% 0.1% 
Close Modal

or Create an Account

Close Modal
Close Modal