Transfer Learning Beyond the Standard Model

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Krishnaraj, Veena, Bayer, Adrian E., Jespersen, Christian Kragh, Melchior, Peter
Format: Preprint
Veröffentlicht: 2025
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866911226171228160
author Krishnaraj, Veena
Bayer, Adrian E.
Jespersen, Christian Kragh
Melchior, Peter
author_facet Krishnaraj, Veena
Bayer, Adrian E.
Jespersen, Christian Kragh
Melchior, Peter
contents Machine learning enables powerful cosmological inference but typically requires many high-fidelity simulations covering many cosmological models. Transfer learning offers a way to reduce the simulation cost by reusing knowledge across models. We show that pre-training on the standard model of cosmology, $Λ$CDM, and fine-tuning on various beyond-$Λ$CDM scenarios -- including massive neutrinos, modified gravity, and primordial non-Gaussianities -- can enable inference with significantly fewer beyond-$Λ$CDM simulations. However, we also show that negative transfer can occur when strong physical degeneracies exist between $Λ$CDM and beyond-$Λ$CDM parameters. We consider various transfer architectures, finding that including bottleneck structures provides the best performance. Our findings illustrate the opportunities and pitfalls of foundation-model approaches in physics: pre-training can accelerate inference, but may also hinder learning new physics.
format Preprint
id arxiv_https___arxiv_org_abs_2510_19168
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Transfer Learning Beyond the Standard Model
Krishnaraj, Veena
Bayer, Adrian E.
Jespersen, Christian Kragh
Melchior, Peter
Cosmology and Nongalactic Astrophysics
Instrumentation and Methods for Astrophysics
Machine Learning
Data Analysis, Statistics and Probability
Machine learning enables powerful cosmological inference but typically requires many high-fidelity simulations covering many cosmological models. Transfer learning offers a way to reduce the simulation cost by reusing knowledge across models. We show that pre-training on the standard model of cosmology, $Λ$CDM, and fine-tuning on various beyond-$Λ$CDM scenarios -- including massive neutrinos, modified gravity, and primordial non-Gaussianities -- can enable inference with significantly fewer beyond-$Λ$CDM simulations. However, we also show that negative transfer can occur when strong physical degeneracies exist between $Λ$CDM and beyond-$Λ$CDM parameters. We consider various transfer architectures, finding that including bottleneck structures provides the best performance. Our findings illustrate the opportunities and pitfalls of foundation-model approaches in physics: pre-training can accelerate inference, but may also hinder learning new physics.
title Transfer Learning Beyond the Standard Model
topic Cosmology and Nongalactic Astrophysics
Instrumentation and Methods for Astrophysics
Machine Learning
Data Analysis, Statistics and Probability
url https://arxiv.org/abs/2510.19168