Difference: 2018-04-04 ( vs. 1)

• ZTF updates
• pipeline updates
• RB updates (new model has been delivered - to be deployed in a day or two)
• scanning based on transient marshal
• zooniverse scanning and planned campaigns
• working on paper sections
• deep learning for asterisks
• other topics.

Attendees: Ashish, Umaa, Charlotte, Frank, Rahul, David, Adam, Tiara, Zach

Apologies: Matthew, Ragnhild

Frank:

• filter exchanger being tried, so alerts from both filters for the first time
• Some issue with cryostat temp.
• SG in place now
• RB classifier being unit tested - should be in today

Umaa: slides

• tv6fv4c3 based on ~5500 - balanced by oversampling bogus
• eliminated some of the position based features (ranr, decnr)
• that makes the false negatives a bit worse - but thats better than having wrong features
• Ashish: take i band features out,
• Frank: also zero point features
• Ashish: This does not include SNR<15, right?
• Umaa: correct
• classstar was tossed out
• Umaa: sentinel values for ss near objects
• Frank: associated objects should have high class-star
• Adam: look for correlation matrix (especially with ss dist near)
• Charlotte: magnr, ssdistnr now improved, so those are helping more with RB

Mar 24 is when we started getting new ref images. That may change things a bit for RB. Now the requirement is refimages with depth 15

Adam: experience with scanning on transient marshal; packet metadata recently fixed

Ashish: Paper volunteering

Umaa: documenting, git hub transition

-- AshishMahabal - 11 Apr 2018