Clarification re. Fixed Weight (FW) implementation

[sukun1045]

The intent of Fixed Weight (FW) and Fixed State (FS) strategies is to weaken the decoder such that the encoder final hidden state can be used as a robust representation for unsupervised skeleton-based action recognition.

In contrast to FW name, in the original implementation, the weights of the decoder are not kept fixed. The main feature of this strategy is that the decoder receives zero inputs such that it reconstructs sequences based on the encoder final state only. A more appropriate name would probably be Fixed Zero (FZ) for this strategy. Thanks to hyk1996 and Yutasq for pointing out and performing detailed analysis.

[Miss-DN]

Hi. So in my opinion, the preprocessing, namely view-invariant transformation is much more important than the Fixed Zero strategy and it is a bit unfair to compare with other approaches. Have you ever try some ablatioin studies on the preprocessing?

[sukun1045]

@Miss-DN Hi, the view-invariant transformation is very important in our case and that's why we put it at the very beginning of the section of Methods. However, I don't think it is unfair to compare with other approaches because, in any approach, the preprocessing is part of the method and all of these preprocessing operations are unsupervised.

[Miss-DN]

@sukun1045 Thank you for the answer.