Simulation-based inference with scattering representations: scattering is all you need

2024/10/11 by Kiyam Lin, Benjamin Joachimi, Lin, Kiyam +3
Medicine · #Cosmology and Nongalactic Astrophysics (astro-ph.CO) #FOS: Computer and information sciences #FOS: Physical sciences #Instrumentation and Methods for Astrophysics (astro-ph.IM) #Machine Learning (cs.LG) #Machine Learning (stat.ML) #Medical Imaging Techniques and Applications

paper · pdf · doi:10.48550/arxiv.2410.11883

openalex publication_date 2024/10/11 · openalex created_date 2024/10/20 · openalex updated_date 2026/07/28

Abstract

We demonstrate the successful use of scattering representations without further compression for simulation-based inference (SBI) with images (i.e. field-level), illustrated with a cosmological case study. Scattering representations provide a highly effective representational space for subsequent learning tasks, although the higher dimensional compressed space introduces challenges. We overcome these through spatial averaging, coupled with more expressive density estimators. Compared to alternative methods, such an approach does not require additional simulations for either training or computing derivatives, is interpretable, and resilient to covariate shift. As expected, we show that a scattering only approach extracts more information than traditional second order summary statistics.

Related

• How to quantify fields or textures? A guide to the scattering transform
• Simulation-based inference has its own Dodelson-Schneider effect (but it knows that it does)
• Simulation-Based Inference: A Practical Guide
• From few to many maps: A fast map-level emulator for extreme augmentation of CMB systematics datasets
• Mitigating Model Misspecification in Simulation-Based Inference for Galaxy Clustering