Thinklab-SJTU/NAR-CO-Solver

Possible solution for Maximum Clique

Pamphlett opened this issue · 2 comments

Pamphlett commented

Hi Runzhong,

Really appreciated your great work!
I'm wondering if it's possible to have a NN solver for the maximum clique problem?
Could you share some insights on this?

rogerwwww commented

Hi Pengyu,

You raised a very interesting point. You can take a look at our recent ICML'23 paper (https://github.com/Thinklab-SJTU/LinSATNet). If you replace the Sinkhorn layer in this repo with LinSATLayer, it might work.

  • 👍1
Pamphlett commented

Thanks for the reference. I'll go and have a try!

  • ❤️1