2019-04-24 Brown Dog Meeting notes

• Developed green index extractor for images and deployed on development instance
  • does not have a bamboo plan - does not have permissions to make one? - get with Rob
  • will then need to go to production
  • Shared docs with Matt Browning and team. Will schedule a meeting next week.
• Met with Matias about dataset question and Box skill in general
  • Dataset is curated subset of DES DR1 (https://des.ncsa.illinois.edu/releases/dr1) with images of galaxies with higher brightness levels
  • It contains about 60,000+ images. The model was trained using about 12,000+ images and currently the Box skill finds similar galaxies from these 12K images.
  • There is metadata associated with each of the galaxies, which might be useful to a researcher. Would need to use a different card to display (transcript card maybe?)
    
    • Example:
      • ID                  RA                 DEC             G                     R                      I                     Y                      Z                      NAME
      • 392371906  59.556006  -31.279512  16.57033157  15.91237926  15.5947876  15.27796268  15.38665581  DESJ035813.4-311646.2
  • Paper about dataset: https://arxiv.org/abs/1801.03181
  • Have not communicated with the student who requested access yet - Nikhar Mamtora
    
    • give him the directory to try it out
    • only give temporary access

not much time to work on docker compose file - have found some Rob files released for Pecan which could be reviewed for info -

focus on bd fiddle, fence, rest of what he already has

persistent volume - yes - mongo, filesystems for polyclot and clowder should be persistent

redis snapshots - so will need it

Tools catalog next

saw test failures a couple hours ago - will look at them

dbpedia is down because their end point is down - hoping it will come back up OR may need to change to a different end point - can we catch this and display the endpoint data of "under maintenance"

good meetings with Pollen team - setting up clowder instance for them to play with

we need to discuss when people try things out - should we just point them to a central instance - clowder demo and maintain properly

learning tensor flow and keras 

checking information on image file - not on pavement files

try sidewalk images first

GPU nodes - can sign up for nodes / make a request for this project

Barbara also has a GPU node as well