Closed tomeichlersmith closed 4 years ago
A simple place to start would be reproducing the gradient BDT results using the new library. This can be done outside of ldmx-sw. Once we have decided which library to use, we can integrate it in.
I should note that we can also integrate multiple libraries but my preference would be to use tensor flow.
We would like to incorporate a computer learning library into ldmx-sw so that studies using Boosted Decision Trees or Neural Networks can be done within ldmx-sw and processors in ldmx-app.
The first goal would be to transition the Ecal Veto BDT to using this incorporated library instead of calling xgboost from within TPython.
Two suggested options are: