Academia.eduAcademia.edu

Figure 1 – uploaded by Jiongzhi Zheng

See full PDFdownloadDownload figure
Figure 1: The choice of x;, y;, xj+1 and y;+ in k-opt.

Figure 1 The choice of x;, y;, xj+1 and y;+ in k-opt.

Related Figures (4)

Table 1. Comparison of VSR-LKH and DRL algorithms on five easy instances (best results in bold).
Table 1. Comparison of VSR-LKH and DRL algorithms on five easy instances (best results in bold).
Figure 2: Q-value evaluation on cumulative gap and runtime (instances are sorted in ascending order of problem size).
Figure 2: Q-value evaluation on cumulative gap and runtime (instances are sorted in ascending order of problem size).
Figure 3: Composite versus single reinforcement methods on cumulative gap and runtime.
Figure 3: Composite versus single reinforcement methods on cumulative gap and runtime.
Table 2. Comparison of VSR-LKH and LKH on some hard instances, best results in bold (see full results in Appendix).
Table 2. Comparison of VSR-LKH and LKH on some hard instances, best results in bold (see full results in Appendix).
Related topics:
Computer ScienceReinforcement Learning
580 California St., Suite 400
San Francisco, CA, 94104
© 2025 Academia. All rights reserved