Optional relabelling of nodes & edges with Annotation.UNKNOWN label

[KristinaUlicna]

If the user decides to do so, relabel all GT annotations from Annotation.UNKNOWN to Annotation.TRUE_NEGATIVE, to increase the number of data points to train the GCN on. Particularly good for exhaustive annotation.

• Resolves Include optional argument to relabel Annotation.UNKNOWN label  #185
• Resolves Print out image / graph node & edge statistics if desired #201
• Resolves Handle montage display with only 2 annotation classes #197
• Resolves Use logging rather than print for user feedback #204

[review-notebook-app]

Check out this pull request on 

See visual diffs & provide feedback on Jupyter Notebooks.

---

Powered by ReviewNB

[crangelsmith]

I'm trying to run the code, but I get a fail with the following error:

File "/Users/crangelsmith/Projects/grace/grace/models/classifier.py", line 72, in forward
    embeddings, p=self.dropout, training=self.training
    ^^^^^^^^^^
UnboundLocalError: cannot access local variable 'embeddings' where it is not associated with a value

I think it is due to the embeddings variable in this line not being defined is the condition in line 66 is not met.

grace/grace/models/classifier.py

Line 69 in a4d98ef

embeddings = x

[KristinaUlicna]

I'm trying to run the code, but I get a fail with the following error:

File "/Users/crangelsmith/Projects/grace/grace/models/classifier.py", line 72, in forward
    embeddings, p=self.dropout, training=self.training
    ^^^^^^^^^^
UnboundLocalError: cannot access local variable 'embeddings' where it is not associated with a value

I think it is due to the embeddings variable in this line not being defined is the condition in line 66 is not met.

grace/grace/models/classifier.py

Line 69 in a4d98ef

embeddings = x

Temporarily fixed now, proper fix for option to train a linear-only classifier upcoming in #183 🚀

[crangelsmith]

I'm trying to run the code, but I get a fail with the following error:

File "/Users/crangelsmith/Projects/grace/grace/models/classifier.py", line 72, in forward
    embeddings, p=self.dropout, training=self.training
    ^^^^^^^^^^
UnboundLocalError: cannot access local variable 'embeddings' where it is not associated with a value

I think it is due to the embeddings variable in this line not being defined is the condition in line 66 is not met.

grace/grace/models/classifier.py

Line 69 in a4d98ef

embeddings = x

Temporarily fixed now, proper fix for option to train a linear-only classifier upcoming in #183 🚀

Cool, it runs now without issues.

[crangelsmith]

I've run the code with the relabelling annotation on and off and seems to run well. My only comment left is on making the relabeling a configurable option for the user, but besides that I think is ready to merge.