CoMatch icon indicating copy to clipboard operation
CoMatch copied to clipboard

Published 20 hours ago • verified publisher icon salesforce

Several Questions

Open myt889 opened this issue 3 years ago • 2 comments

Hi, Thank you for your great work, I am new to this area, I have the following questions: 1、How much improvements can be made by the branch of Graph-based contrastive learning? I can not see that ablation experiments in the paper. I'm interested in this part and would also like to know the performance improvements that might be useful for my task. 2、Does the Co-match method fit for multi-label classification tasks？

Once again, thank you for your nice work and clean code! Looking forward to your reply. Best regards! Tan

myt889 avatar Feb 26 '22 06:02 myt889

Thanks for your questions!

  1. The graph-based contrastive loss gives a huge improvement. As shown in Figure 4b, reducing the contrastive loss weight from 10 to 1 leads to a big drop, and the performance further degrades if the weight is 0.
  2. Yes the method should be able to easily extend to multi-label classification.

LiJunnan1992 avatar Feb 26 '22 10:02 LiJunnan1992

Ok. Thank you for your quick reply. I will try your method, and hope you can offer guidence if needed. Once again, I really appreciate your works.

Best regards!

myt889 avatar Feb 26 '22 14:02 myt889