hep-ml-lab 0.2.0

An end-to-end framework for applying machine learning to high-energy physics research.

HEP ML Lab (HML)

❗ This framework is currently undergoing rapid iteration. Any comments and suggestions are welcome.

Introduction

HEP ML Lab (HML) is an end-to-end framework for applying machine learning (ML) to high energy physics (HEP) research. It provides a set of interfaces for data generation, model training and evaluation. It is designed to be modular and extensible so that you can easily customize it for your own research.

Module overview

Here is a brief overview of the modules in HML:

• hml.generators: API of Madgraph5 for simulating colliding events;
• hml.theories: Particle physics models;
• hml.observables: General observables in jet physics;
• hml.representations: Different data structure used to represent an event;
• hml.datasets: Existing datasets and helper classes for creating new datasets;
• hml.methods: Cuts, trees and networks for classification;
• hml.metrics: Metrics used in classical signal vs background analysis;