The loss of STSGCN is irregularly large #311
Merged
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
I am currently testing the STSGCN and found the loss is irregularly large (>1e10).
I read the code and found the output is accumulated after the graph convolution:
Bigscity-LibCity/libcity/model/traffic_flow_prediction/STSGCN.py
Line 50 in 85d1351
I checked the paper and code and thought they use the distance inverse as the adjacent matrix, rather than the Gauss kernel (https://github.com/Davidham3/STSGCN/blob/master/utils.py#L105).
I added a new function in traffic_state_dataset to calculate the distance inverse for STSGCN and the loss now looks reasonable
(PEMSD7(M), 12 prediction steps):