Attendees: Ashot, Chao G, Chao P, Maxime, Weizhi, Xinzhan, Eugene, Nilanga, Dipangkar, Haiyan, Xuefei

Weizhi:
- calibration finished for all data 1GeV and 2GeV
- working on resolution vs. E
- need to check one by one for some problematic modules in LG region
- TASK: need to compare with snake run resolution (Maxime), especially top and right LG region

Xuefei:
- Training in 2 different theta region:
  - 2<theta<5 degree: satisfactory training
  - >5 degree: not enough statistics from cosmic runs, need to use empty runs (introduce a bias)
- Quantification of training and matching efficiencies (and bias)
- possible improvement for >5 degree region with more sophisticated tools

Xinzhan:
- some obvious cosmic events are not rejected with the neural network
- TASK: select these obvious events with some other cuts and train the neural network specifically with it
(also possible to do cosmic rejection in 2 separate steps: obvious cuts (cluster profile, excentricity) and neural network)