██  ██████ ███████ ███    ██ ███████ ████████
██ ██      ██      ████   ██ ██         ██   
██ ██      █████   ██ ██  ██ █████      ██   
██ ██      ██      ██  ██ ██ ██         ██   
██  ██████ ███████ ██   ████ ███████    ██   

A deep learning driven library for high energy physics++

First steps

See autogenerated docs:

https://mieskolainen.github.io/icenet

Start with the installation:

docs/source/notes/installation.rst

For end-to-end deep learning examples, see e.g. github actions (CI) workflows under

.github/workflows

AI-Tech Docs

Introduction talk (2023):

docs/pdf/CMS_ML_forum_Mieskolainen_080223.pdf

Conditional AI reweighting (2025):

configs/zee/NOTES_reweight.md

ICEBOOST = xgboost + torch autograd (2025):

configs/zee/NOTES_iceboost.md

Reference

If you use this work in your research -- especially if you find algorithms, their application or ideas novel, please include a citation:

@software{icenet,
  author  = "{Mikael Mieskolainen}",
  title   = "ICENET: a deep learning library for HEP",
  url     = "https://github.com/mieskolainen/icenet",
  version = {X.Y.Z},
  date    = {2025-XX-YY},
}

Mikael Mieskolainen, 2025
[email protected]