Contrastive Learning for Robust Representations of Neutrino Data

TL;DR

This paper explores how contrastive learning can improve the robustness and transferability of models trained on simulated neutrino data, enhancing their performance on real detector data in neutrino physics.

Contribution

It introduces the application of contrastive learning to neutrino data analysis, demonstrating its advantages over other domain adaptation methods.

Findings

Contrastive learning improves model robustness to real data variations.

Enhanced transferability of features from simulated to real neutrino data.

Comparison shows contrastive learning outperforms traditional domain adaptation techniques.

Abstract

In neutrino physics, analyses often depend on large simulated datasets, making it essential for models to generalise effectively to real-world detector data. Contrastive learning, a well-established technique in deep learning, offers a promising solution to this challenge. By applying controlled data augmentations to simulated data, contrastive learning enables the extraction of robust and transferable features. This improves the ability of models trained on simulations to adapt to real experimental data distributions. In this paper, we investigate the application of contrastive learning methods in the context of neutrino physics. Through a combination of empirical evaluations and theoretical insights, we demonstrate how contrastive learning enhances model performance and adaptability. Additionally, we compare it to other domain adaptation techniques, highlighting the unique advantages of contrastive learning for this field.