not reproducing results

[GonyRosenman]

hello,
thanks for sharing your code.
i'm trying to reproduce results but not getting similar accuracy.

steps i took:

1. download the folder 'hcp_tc_npy_22' from the google drive link that you supplied.
2. run the preprocessing.py script after configuring the folders.
3. run the 'run_training.py' script.

i made minor changes such as configuring 'adj_matrix.npy' isntead of 'adj_matrix_qingyu.npy'. (since this adj_matrix is generated automatically in the preprocessing.py script)

the accuracy doesn't go over 55%. what can possibly explain this?

[sgadgil6]

Can you try running the run_training.py with import net.st_gcn instead of import net.st_gcn_lstm

[GonyRosenman]

hello again,
changing to net.st_gcn improved results up to 75-77% accuracy (:
is it possible that the adjacency matrix 'adj_matrix_qingyu.npy' is causing better performance as seen in your paper?
maybe you can give details about how to reproduce this matrix if so.
thanks!

[GMLB1997]

I meet the same problem. The results in the paper cannot be reproduced. Only 75-77% accuracy can be achieved