Enhancing Cosmological Model Selection with Interpretable Machine Learning

Abstract

We propose a novel approach using neural networks (NNs) to differentiate between cosmological models, and implemented LIME as an interpretability approach to identify the key features influencing our model's decisions. We show the potential of NNs to enhance the extraction of meaningful information from cosmological large-scale structure data, based on current galaxy-clustering survey specifications, for the cosmological constant and cold dark matter ($\Lambda$CDM) model and the Hu-Sawicki $f(R)$ model. We find that the NN can successfully distinguish between $\Lambda$CDM and the $f(R)$ models, by predicting the correct model with approximately $97\%$ overall accuracy, thus demonstrating that NNs can maximize the potential of current and next generation surveys to probe for deviations from general relativity.

7 pages, 6 figures, changes match published version

References in corpus (54)

• f(R) Theories Of Gravity
• f(R) theories
• Extended Theories of Gravity
• The clustering of galaxies in the completed SDSS-III Baryon Oscillation Spectroscopic Survey: cosmological analysis of the DR12 galaxy sample
• The clustering of galaxies in the SDSS-III Baryon Oscillation Spectroscopic Survey: Baryon Acoustic Oscillations in the Data Release 10 and 11 galaxy samples
• Models of f(R) Cosmic Acceleration that Evade Solar-System Tests
• The Simons Observatory: Science goals and forecasts
• The Completed SDSS-IV extended Baryon Oscillation Spectroscopic Survey: Cosmological Implications from two Decades of Spectroscopic Surveys at the Apache Point observatory
• Dark Energy Survey Year 3 Results: Cosmological Constraints from Galaxy Clustering and Weak Lensing
• Challenges for $\Lambda$CDM: An update
• Baryon Acoustic Oscillations in the Lyα forest of BOSS DR11 quasars
• Galaxy Clusters Discovered via the Sunyaev-Zel'dovich Effect in the 2500-square-degree SPT-SZ survey
• The Atacama Cosmology Telescope: DR4 Maps and Cosmological Parameters
• On the Robustness of the Acoustic Scale in the Low-Redshift Clustering of Matter
• A Parameterized Post-Friedmann Framework for Modified Gravity
• Deep Learning and its Application to LHC Physics
• Cluster Cosmology Constraints from the 2500 deg${}^2$ SPT-SZ Survey: Inclusion of Weak Gravitational Lensing Data from Magellan and the Hubble Space Telescope
• Jet Substructure at the Large Hadron Collider: A Review of Recent Advances in Theory and Machine Learning
• Deep-learning Top Taggers or The End of QCD?
• Searching for modified growth patterns with tomographic surveys
• Testing gravity with CAMB and CosmoMC
• Dark Energy versus Modified Gravity
• DESI 2024: Reconstructing Dark Energy using Crossing Statistics with DESI DR1 BAO data
• Crossing the Phantom Divide
• Quintessential interpretation of the evolving dark energy in light of DESI
• Observational constraints on viable f(R) parametrizations with geometrical and dynamical probes
• Dark Energy Survey Year 3 Results: A 2.7% measurement of Baryon Acoustic Oscillation distance scale at redshift 0.835
• $Euclid$ preparation: XV. Forecasting cosmological constraints for the $Euclid$ and CMB joint analysis
• MGCAMB with massive neutrinos and dynamical dark energy
• Galaxy morphology rules out astrophysically relevant Hu-Sawicki $f(R)$ gravity
• Quantifying Scalar Field Dynamics with DESI 2024 Y1 BAO measurements
• The miniJPAS survey: star-galaxy classification using machine learning
• Fast and Credible Likelihood-Free Cosmology with Truncated Marginal Neural Ratio Estimation
• Star formation rates and stellar masses from machine learning
• Euclid: impact of nonlinear prescriptions on cosmological parameter estimation from weak lensing cosmic shear
• The Young Supernova Experiment Data Release 1 (YSE DR1): Light Curves and Photometric Classification of 1975 Supernovae
• Baryon Acoustic Oscillation Theory and Modelling Systematics for the DESI 2024 results
• The Dark Energy Survey 5-year photometrically identified Type Ia Supernovae
• Accuracy of the growth index in the presence of dark energy perturbations
• New MGCAMB tests of gravity with CosmoMC and Cobaya
• The Dark Energy Survey Supernova Program: Cosmological biases from supernova photometric classification
• Euclid preparation: XXII. Selection of Quiescent Galaxies from Mock Photometry using Machine Learning
• Euclid preparation. XLI. Galaxy power spectrum modelling in real space
• Does jackknife scale really matter for accurate large-scale structure covariances?
• Euclid preparation. XLIII. Measuring detailed galaxy morphologies for Euclid with machine learning
• Evidence Networks: simple losses for fast, amortized, neural Bayesian model comparison
• Euclid Preparation XXXIII. Characterization of convolutional neural networks for the identification of galaxy-galaxy strong lensing events
• Euclid: Identification of asteroid streaks in simulated images using deep learning
• Exploring Supernova Gravitational Waves with Machine Learning
• Bayesian evidence estimation from posterior samples with normalizing flows
• Accurate Computation of the Screening of Scalar Fifth Forces in Galaxies
• Euclid: Identifying the reddest high-redshift galaxies in the Euclid Deep Fields with gradient-boosted trees
• Faster Bayesian inference with neural network bundles and new results for $f(R)$ models
• Euclid: Testing photometric selection of emission-line galaxy targets

Cited by in corpus (4)

• Interpreting anomaly detection of SDSS spectra
• Neural Networks for cosmological model selection and feature importance using Cosmic Microwave Background data
• Conditional variational autoencoders for cosmological model discrimination and anomaly detection in cosmic microwave background power spectra
• Interpretability of deep-learning methods applied to large-scale structure surveys