BIP: Boost Invariant Polynomials for Efficient Jet Tagging

TL;DR

This paper introduces BIP, a boost-invariant polynomial framework for jet tagging that is both highly accurate and significantly more efficient than existing deep learning methods in high energy physics.

Contribution

The paper proposes a new invariant polynomial approach for jet tagging that improves computational efficiency and interpretability over traditional deep learning models.

Findings

Achieves high accuracy on jet tagging benchmarks.

Orders of magnitude faster training and evaluation.

Applicable to both supervised and unsupervised schemes.

Abstract

Deep Learning approaches are becoming the go-to methods for data analysis in High Energy Physics (HEP). Nonetheless, most physics-inspired modern architectures are computationally inefficient and lack interpretability. This is especially the case with jet tagging algorithms, where computational efficiency is crucial considering the large amounts of data produced by modern particle detectors. In this work, we present a novel, versatile and transparent framework for jet representation; invariant to Lorentz group boosts, which achieves high accuracy on jet tagging benchmarks while being orders of magnitudes faster to train and evaluate than other modern approaches for both supervised and unsupervised schemes.