Beyond mirkwood: Enhancing SED Modeling with Conformal Predictions

TL;DR

The paper tackles uncertainties in spectral energy distribution fitting caused by star formation histories and dust attenuation by introducing a flexible ML framework that accepts any sklearn-compatible model and employs conformalized quantile regression for reliable uncertainty. Using CatBoost as the base predictor and MAPIE for interval calibration, the approach yields improved coverage and narrower prediction intervals compared with Prospector and mirkwood across simulations. The methodology is validated on 35-band photometry from IllustrisTNG, Eagle, and Simba, with a sequential prediction scheme that propagates uncertainties through to all four galaxy properties. This work provides a versatile, uncertainty-aware tool for inferring galaxy properties from observational data, with potential applicability to upcoming large surveys.

Abstract

Traditional spectral energy distribution (SED) fitting techniques face uncertainties due to assumptions in star formation histories and dust attenuation curves. We propose an advanced machine learning-based approach that enhances flexibility and uncertainty quantification in SED fitting. Unlike the fixed NGBoost model used in mirkwood, our approach allows for any sklearn-compatible model, including deterministic models. We incorporate conformalized quantile regression to convert point predictions into error bars, enhancing interpretability and reliability. Using CatBoost as the base predictor, we compare results with and without conformal prediction, demonstrating improved performance using metrics such as coverage and interval width. Our method offers a more versatile and accurate tool for deriving galaxy physical properties from observational data.