@codybontecou thanks for the comment! Yup I will add the arxiv link for it.

Thanks for the comment! Yes you are right on the role of the filter mechanism. Essentially, - In GNN design, knowing graph similarity helps predict if a GNN that works on one graph will perform well on another. - The challenge is finding which topological similarity aligns with model performance (empirical similarity). - Correlation measures how well topological similarity (based on a feature) predicts empirical performance similarity. - High correlation means the feature is a good predictor of GNN performance transfer between graphs. - Low or negative correlation means the feature is not reliable for transferring GNN knowledge. - Statistical and empirical indicators alone don’t consistently explain how graph similarities affect GNN performance across datasets. - The adaptive filtering mechanism dynamically selects the most relevant graph features based on their actual impact on GNN performance. - For each benchmark graph, the mechanism calculates the correlation between each feature and empirical performance. - The mechanism then computes the average correlation for each feature across all graphs. - Based on these averages, the filter selects the top-N most influential features for GNN design or selection. I will try to make my commentary in more lay-man terms. As for the background, I am still trying to find a program that can remove the background without breaking my computer cpu haha. If you have any suggestions on that, it would be nice.

Thanks man. Good luck to you as well!

oh dear, I just noticed it. Thanks for bringing this up. Let me check if I can recover the uncorrupted video...

😂

Thanks!

I sure hope it does!

Thanks!

I’m working on a paper related to segmentation with transformers and I could really use some help , can you please me give me your email contact, so that I can reach out to you?

That will be coming soon! Stay tuned~~~

Vision transformers for semantic segmentation is here! :P ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-rHcx0iFWP6s.html

Yea I am using a laptop mic. thats why it kinda bad :P

Thanks for watching pal!