Improving Parametric Neural Networks for High-Energy Physics (and Beyond)

TL;DR

This paper enhances parametric neural networks for high-energy physics by analyzing their peculiarities, proposing a new architecture called AffinePNN, and validating improvements through extensive experiments on the HEPMASS dataset.

Contribution

It introduces an alternative parametrization scheme, the AffinePNN, along with improved training procedures and comprehensive evaluation on real-world datasets.

Findings

AffinePNN outperforms previous pNN models in classification accuracy.

Balanced training improves model stability and performance.

The new approach demonstrates strong interpolation capabilities.

Abstract

Signal-background classification is a central problem in High-Energy Physics (HEP), that plays a major role for the discovery of new fundamental particles. A recent method -- the Parametric Neural Network (pNN) -- leverages multiple signal mass hypotheses as an additional input feature to effectively replace a whole set of individual classifiers, each providing (in principle) the best response for the corresponding mass hypothesis. In this work we aim at deepening the understanding of pNNs in light of real-world usage. We discovered several peculiarities of parametric networks, providing intuition, metrics, and guidelines to them. We further propose an alternative parametrization scheme, resulting in a new parametrized neural network architecture: the AffinePNN; along with many other generally applicable improvements, like the balanced training procedure. Finally, we extensively and empirically evaluate our models on the HEPMASS dataset, along its imbalanced version (called HEPMASS-IMB) we provide here for the first time, to further validate our approach. Provided results are in terms of the impact of the proposed design decisions, classification performance, and interpolation capability, as well.