Which feature extractor do you use to get the result in table2?
2020213484 opened this issue · 6 comments
It seems that the feature extractor is not claimed for the results in table2. Should it be graph wavenet (GWN)?
I would be grateful if you could provide clarification on this problem.
In Table 2, we did the ablation study based on GRU feature extractor.
Does this imply that the optimal result for ST-GFSL was achieved using GRU as the feature extractor?
Nope, ST-GFSL is a universal framework, which can be combined with any feature extractor.
I'm sorry that I fail to clearly clarify my confuse. Of course I know that ST-GFSL is a universal framework. What I mean is which feature extractor is combined with to get the result for ST-GFSL on Table 2?
I believe that you choose the proper feature extractor to make the ST-GFSL achieve the best result.
In Table 2, we directly use GRU model as feature extractor. ST-GFSL shows consistently performance improvement in most feature extractors.
Thank you for your patient reply !