List view

• WP1: Create a minimum viable product

  Please help verify the following points: - SL with VIHO - [x] Environment set up - [ ] Att-mil training - [x] 3D-CNN training - SL with UCHICAGO - [x] Environment set up - [ ] Att-mil training - SL with UKA - [x] Environment set up - [ ] Att-mil training - [x] 3D-CNN training - SL with Radboud - [x] Environment set up - [ ] Att-mil training - [x] 3D-CNN training - [x] Unnecessary files are deleted and added to .gitignore - [x] Credentials are not exposed - [x] VPN username and password input - [x] scp command requires the password for our swarm account - [ ] Swarm learning email account is used instead of personal ones - [x] HPE login - [x] VPN account - [ ] Docker image storage - [x] Google Drive for models and dataset - [x] Necessary steps are well documented - [x] replace all "set -eux" in shell scripts to "set -eu" after all the validations

  No due date

  •1/2 issues closed
  50% complete1 open1 closed

• Swarm learning ducumentation and trouble-shooting

  Any problems encountered when using HPE platform and worth investigation.

  No due date

  •7/21 issues closed
  33% complete14 open7 closed

• Implement radiology data training on SL

  Use the attMIL approach for MRI which Marta used for the GBM for our breast dataset. At this point, we only have one sequence (T2w images). it's an adapted version of marugoto.

  No due date

  •6/7 issues closed
  85% complete1 open6 closed