Interpreting deep learning-based stellar mass estimation via causal analysis and mutual information decomposition

Abstract

End-to-end deep learning models fed with multi-band galaxy images are powerful data-driven tools used to estimate galaxy physical properties in the absence of spectroscopy. However, due to a lack of interpretability and the associational nature of such models, it is difficult to understand how the information that is included in addition to integrated photometry (e.g., morphology) contributes to the estimation task. Improving our understanding in this field would enable further advances into unraveling the physical connections among galaxy properties and optimizing data exploitation. Therefore, our work is aimed at interpreting the deep learning-based estimation of stellar mass via two interpretability techniques: causal analysis and mutual information decomposition. The former reveals the causal paths between multiple variables beyond nondirectional statistical associations, while the latter quantifies the multicomponent contributions (i.e., redundant, unique, and synergistic) of different input data to the stellar mass estimation. Using data from the Sloan Digital Sky Survey (SDSS) and the Wide-field Infrared Survey Explorer (WISE), we obtained meaningful results that provide physical interpretations for image-based models. Our work demonstrates the gains from combining deep learning with interpretability techniques, and holds promise in promoting more data-driven astrophysical research (e.g., astrophysical parameter estimations and investigations on complex multivariate physical processes).

Accepted at Astronomy & Astrophysics; 23 + 12 pages; 8 + 16 figures

References in corpus (38)

• The Wide-field Infrared Survey Explorer (WISE): Mission Description and Initial On-orbit Performance
• LSST: from Science Drivers to Reference Design and Anticipated Data Products
• The physical properties of star forming galaxies in the low redshift universe
• The Eleventh and Twelfth Data Releases of the Sloan Digital Sky Survey: Final Data from SDSS-III
• Stellar Masses and Star Formation Histories for 10^5 Galaxies from the Sloan Digital Sky Survey
• The propagation of uncertainties in stellar population synthesis modeling I: The relevance of uncertain aspects of stellar evolution and the IMF to the derived physical properties of galaxies
• UV Star Formation Rates in the Local Universe
• Explaining NonLinear Classification Decisions with Deep Taylor Decomposition
• NYU-VAGC: a galaxy catalog based on new public surveys
• Photometric redshifts from SDSS images using a Convolutional Neural Network
• The Intrinsic Properties of SDSS Galaxies
• Decomposing causality into its synergistic, unique, and redundant components
• An information theoretic framework for classifying exoplanetary system architectures
• Feature importance for machine learning redshifts applied to SDSS galaxies
• Morpho-z: improving photometric redshifts with galaxy morphology
• Star formation rates and stellar masses from machine learning
• How Many Elements Matter?
• New Approaches To Photometric Redshift Prediction Via Gaussian Process Regression In The Sloan Digital Sky Survey
• A machine learning approach to galaxy properties: joint redshift-stellar mass probability distributions with Random Forest
• Using convolutional neural networks to predict galaxy metallicity from three-color images
• Analysis of a Custom Support Vector Machine for Photometric Redshift Estimation and the Inclusion of Galaxy Shape Information
• Return of the features. Efficient feature selection and interpretation for photometric redshifts
• A Causal, Data-Driven Approach to Modeling the Kepler Data
• Star Formation Rates for photometric samples of galaxies using machine learning methods
• Predicting star formation properties of galaxies using deep learning
• COSMOS2020: Manifold Learning to Estimate Physical Parameters in Large Galaxy Surveys
• The Efficacy of Galaxy Shape Parameters in Photometric Redshift Estimation: A Neural Network Approach
• Euclid preparation: XXII. Selection of Quiescent Galaxies from Mock Photometry using Machine Learning
• Probing large scale homogeneity and periodicity in the LRG distribution using Shannon entropy
• Effect of Inclination of Galaxies on Photometric Redshift
• On the complexity and the information content of cosmic structures
• How to measure metallicity from five-band photometry with supervised machine learning algorithms
• Morphology-assisted galaxy mass-to-light predictions using deep learning
• Galaxy stellar and total mass estimation using machine learning
• Relating information entropy and mass variance to measure bias and non-Gaussianity
• Causal Discovery in Astrophysics: Unraveling Supermassive Black Hole and Galaxy Coevolution
• Planetary Complexity Revealed by the Joint Differential Entropy of Eigencolours
• CLAP. I. Resolving miscalibration for deep learning-based galaxy photometric redshift estimation