ccs-lab / project_model_based_fmri Goto Github PK

Because My pull request is merged by neurosynth, mb-mvpa doesn't use neurosynth as submodule.
Please remove them.

If mb-mvpa doesn't plan to support classification, I recommend combining regressor.py and model.py.
If not, create classifier.py and use the model.py to wrap it.
It is more beneficial for maintenance and readability

Pytorch (facebook) comments style should be changed to google style.
google style guide reference: https://google.github.io/styleguide/pyguide.html

Intentionally left in the issue page.

How about ROI mask automatically from Neurosynth library instead of downloading it from Neurosynth web site?
I uploaded pull request, can you check if it can be used as a submodule?

When I perform penalized linear regression with piva2019 example data, the following error will occur. Please solve it..

WARNING:tensorflow:Can save best model only with val_loss available, skipping.
WARNING:tensorflow:Early stopping conditioned on metric `val_loss` which is not available. Available metrics are: loss

There is a problem with custom masking.
Dimension of custom masking result (a.k.a. m_true, related to threshold..?) is 169 fixed in both two examples.. (I thought this is bug..)
Please fix this problem.. @cheoljun95

In multicore environment, shutdown occurs intermittently when ncore >=4 and nthread >=4.
If we want to use more cores and threads, each process and thread must be changed by reducing the memory they have in common.

The variable is not defined.. Where should I check..?

I think that the problem is conditional indexing error when boolean value put in pandas dataframe.. (e.g. df[True])

Same as the title. In addition to the examples, I am working on the tutorial with a more detailed explanation. I think the full size of data is affordable.