Re-envisioning Euclid Galaxy Morphology: Identifying and Interpreting Features with Sparse Autoencoders

2025/10/27 by Wu, John F., Walmsley, Michael
#FOS: Computer and information sciences #FOS: Physical sciences #Instrumentation and Methods for Astrophysics (astro-ph.IM) #Machine Learning (cs.LG)

paper · doi:10.48550/arxiv.2510.23749

Abstract

Sparse Autoencoders (SAEs) can efficiently identify candidate monosemantic features from pretrained neural networks for galaxy morphology. We demonstrate this on Euclid Q1 images using both supervised (Zoobot) and new self-supervised (MAE) models. Our publicly released MAE achieves superhuman image reconstruction performance. While a Principal Component Analysis (PCA) on the supervised model primarily identifies features already aligned with the Galaxy Zoo decision tree, SAEs can identify interpretable features outside of this framework. SAE features also show stronger alignment than PCA with Galaxy Zoo labels. Although challenges in interpretability remain, SAEs provide a powerful engine for discovering astrophysical phenomena beyond the confines of human-defined classification.

Citations

• Position: Use Sparse Autoencoders to Discover Unknowns
• Identifying Astrophysical Anomalies in 99.6 Million Cutouts from the Hubble Legacy Archive Using AnomalyMatch
• Learning Multi-Level Features with Matryoshka Sparse Autoencoders
• SAEBench: A Comprehensive Benchmark for Sparse Autoencoders in Language Model Interpretability
• Insights on Galaxy Evolution from Interpretable Sparse Feature Networks
• Decoding Dark Matter: Specialized Sparse Autoencoders for Interpreting Rare Concepts in Foundation Models
• A is for Absorption: Studying Feature Splitting and Absorption in Sparse Autoencoders
• AstroMAE: Redshift Prediction Using a Masked Autoencoder with a Novel Fine-Tuning Architecture
• Disentangling Dense Embeddings with Sparse Autoencoders
• Sparse Autoencoders Find Highly Interpretable Features in Language Models
• Scale-MAE: A Scale-Aware Masked Autoencoder for Multiscale Geospatial Representation Learning
• Masked Autoencoders Are Scalable Vision Learners
• Do Self-Supervised and Supervised Methods Learn Similar Visual Representations?
• Shortcut learning in deep neural networks
• Wide-Field InfrarRed Survey Telescope-Astrophysics Focused Telescope Assets WFIRST-AFTA 2015 Report
• k-Sparse Autoencoders

Related

• Morphological classification of astronomical images with limited labelling
• Semi-supervised Learning of Galaxy Morphology using Equivariant Transformer Variational Autoencoders
• Deciphering galaxy images using machine vision -- Combining variational autoencoder and principal component analysis for feature extraction
• Similar image retrieval using Autoencoder. I. Automatic morphology classification of galaxies
• Visualizing the Hidden Features of Galaxy Morphology with Machine Learning