Table 2 Pairwise comparison of TN-GEO against each of the SOTA optimizers

From: Enhancing combinatorial optimization with classical and quantum generative models

TN-GEO vs Other:

GRASP

ABCFEIT

AAG

VNSQP

ABC-HP

NADE-GEO

Wins(+)

12

10

11

11

8

8

Loss( − )

12

9

11

12

16

18

Ties

11

16

13

12

11

9

(Wins+Ties)/Total

66%

74%

69%

66%

54%

77%

Asymptotic significance (p)

0.247

0.888

0.363

0.594

0.110

0.003

Decision

Retain

Retain

Retain

Retain

Retain

Reject

  1. The asymptotic significance is part of the Wilcoxon signed-rank test results. The null hypothesis that the performance of the two algorithms is the same is tested at the 95% confidence level (significance level: α = . 05). Results show that TN-GEO is on par with all the SOTA algorithms. We also report the count for TN-GEO wins, losses, and ties, compared to each of the other algorithms.
Back to article page