The difference in accuracy of reproduction.

[ZziTaiLeo]

Hi， I just download the data.zip from gaitgraph for casia-b_poses_train_valid.csv and casia-b_valid.csv, then use experiments/train.sh to train model, but the accuracy looks have a bit difference from paper. I have not adjusted any hyperparameters or network structure. How can I adjust it to reproduce this accuracy.
Looking forward to your early reply! Tks！

[exitudio]

Hi, Thank you for being interested in our work.
the variance we're seeing is likely because the CASIA-B dataset is relatively small, and the inclusion of bag (BG) and coat (CL) data adds some noise.
We ran the experiment 8 times and reported the best result. Note that, this pattern is not unique to our model—other methods, like GaitGraph, exhibit similar behavior.

[ZziTaiLeo]

Thank you for your prompt reply! I only saw it now. Although you said you had the best results in eight experiments, I still cannot accept the current results and cannot find a corresponding solution to solve it. Are you downloading the processed skeleton dataset directly?

[exitudio]

Yes. We use the preprocessed data from GaitGraph.