4TOPS

ZhangYu: 

1. Implementing b-tagging scale factors, running jobs, waiting to see if everything is consistent
2. Fabio and Huiling should implement these SFs as well (possibly using exactly the same code, i.e., Fabio's function which has been modified by ZhangYu to fit our needs)

Huiling: 

1. Seing weird things in significance plots outputed by TMVA; ZhangYu's method to plot the significance looks stabler and more reliable. Maybe stick to it? TMVA macros have already proven to be buggy...
2. Hideki's suggestion: put error bars in the significance plot
3. Hongbo's suggestion: probably better to have just one BDT for 1tau1L instead of two separate BDTs for 1tau1e and 1tau1mu 

Fabio:

1. Trigger studies look reasonable, we may consider to have a dedicated preselection for 1tau3L and 2tau2L requiring >=2 jets instead of >=4 jets
2. Moving now to find a QCD-enriched control region for the 1tau0L category

TTBAR RESONANCE

We set up weekly meetings on Wednesday afternoons/nights (4:00pm CERN time, 10:00pm Beijing time). Weekly meetings should enhance our progress rate.
Currently udnerstanding how to import a BEST model in the CMSSW framework (i.e., load a TensorFlow .pb file). We have some code to take inspiration from, should not be super difficult.