Fast End-to-End Framework for Cosmological Parameter Inference from CMB Data Using Machine Learning

TL;DR

This work tackles the challenge of extracting cosmological parameters from CMB data in the presence of strong foregrounds by introducing a fast, end-to-end pipeline that combines Analytical Blind Separation (ABS) for component separation with neural-network–based inference trained entirely on simulations. The approach is validated on simulated LiteBIRD and PICO skies, focusing on the optical depth $τ$ and tensor-to-scalar ratio $r$, and achieves competitive 1σ uncertainties (approximately 0.0035–0.0030 for $τ$ and 0.0056–0.0015 for $r$) with recovered values consistent with inputs across most of the parameter space. By integrating ABS with simulation-based NN training, the method offers a computationally efficient alternative to traditional likelihood analyses and scales to large simulation ensembles for future CMB missions. The results demonstrate robust spectral recovery and accurate parameter inference, highlighting the practical impact for forecasting and optimizing upcoming CMB polarization experiments such as LiteBIRD and PICO.

Abstract

Precise estimation of cosmological parameters from the cosmic microwave background (CMB) remains a central goal of modern cosmology and a key test of inflationary physics. However, this task is fundamentally limited by strong foreground contamination, primarily from Galactic emissions, which obscure the faint CMB B-mode polarization signal. In this Letter, we introduce a fast, simulation-based, end to end pipeline that integrates a robust component separation technique with machine-learning, leading to cosmological parameter estimation. Our approach combines the Analytical Blind Separation (ABS) method for foreground removal with a neural network (NN) framework optimized to extract cosmological parameters directly from full-sky simulations. We assess the performance of this methodology for the forthcoming LiteBIRD and PICO satellite missions, designed to detect CMB B modes with unprecedented sensitivity. Applying the pipeline to realistic sky simulations, we obtain 1 sigma errors of 0.0035 (LiteBIRD) and 0.0030 (PICO) for the optical depth tau, and 0.005 (LiteBIRD) and 0.0014 (PICO) for the tensor-to-scalar ratio, r. In all cases, the recovered parameters are consistent with input values within 1 sigma across most of the parameter space. Results for LiteBIRD are in excellent agreement with the latest forecasts from the collaboration. Our findings establish this combined ABS-NN pipeline as a competitive, accurate, and computationally efficient alternative for cosmological parameter inference, offering a powerful framework for forthcoming CMB experiments.

Figures (2)

• Figure 1: Angular power spectra of the input clean CMB simulations (left) and the spectra recovered by ABS from LiteBIRD- (middle) and PICO-like (right) simulations. ABS spectra include the effect of the corresponding largest beam (first row of Tables \ref{['tab:exp_litebird']} and \ref{['tab:exp_PICO']}). The green and blue regions represent the noise levels (frequency dependent) for LiteBIRD and PICO cases, respectively.
• Figure 2: Predicted versus true cosmological parameters ($r$ and $\tau$) for LiteBIRD-like (blue) and PICO-like (orange) instruments. Dots and error bars show the mean and standard deviation from 10 simulations for each of the 200 cosmologies of the test set. The black dotted and dashed lines denote the linear fits to LiteBIRD and PICO results, while the thin diagonal indicates the identity $y^{Pred}=y^{True}$. The corresponding root-mean-square errors are also shown. Bottom panels display the statistical significance of the predictions as a function of the true values, with the dashed line marking the 1$\sigma$ level.