OCCAMS: A Multi-Task Learning-Based Investigation of the CMB Power Spectra
Abstract
How many latent variables are needed to fit the cosmic microwave background (CMB) temperature and polarisation power spectra when a neural network is given no cosmological model, and to what extent do they map onto the standard cosmological parameters? We use multi-task learning with variational autoencoders trained on CLASS-generated spectra (TT, TE, EE) over 2≤ℓ≤2500. No cosmological model is built into the network, and the number of latent variables, L, is varied over L=5, $6$ and $7$. Their fits are compared to the Planck 2018 data using χ̂2, AIC, BIC and Bayesian evidence. Their correspondence with cosmological parameters is measured with normalised mutual information post-training. The loss weights are set within a single training run by uncertainty weighting, with the weight of the disentanglement term either fixed (semi-automated, SA) or learned (completely automated, CA). The reconstruction loss is split over multipole ranges in three configurations that separate the roles of temperature and polarisation and of the low- and high-ℓ ranges. In four of the six combinations of configuration and strategy, the fit improves from L=5 to L=6 and not beyond, and L=6 is the best, or close to the best, in five of the six cases. So, six latent variables are needed, as many as the free parameters of the standard, model-dependent approach. SA gives clearer disentanglement, while CA needs no a priori choice of the disentanglement weight and has the higher evidence at L=5 and $6$: the two strategies are points on a trade-off between interpretability and fit to the data. Notably, the latent variables for τ and As share almost no information in any model, consistent with large-scale polarisation separating the two, as in standard analyses. A deep-learning approach can thus give a compact, data-driven parametrisation of the CMB spectra.
Related
- A consistent, physical, and analytical model for cosmic microwave background observables of reionisation. I. Derivation, constraining power, and detectability of the auto spectra
- Investigation of high-quality spectra of a sample of high-radial-velocity stars
- Primordial Black Hole Abundances and Scalar Induced Gravitational Waves from Finite-Width Power Spectra in a Stiff Thermal History
- Detectability of 21-cm Power Spectra from the Epoch of Reionization using Modern Radio Telescopes
- Blast.jl: Differentiable Non-Limber Power Spectra for Joint Clustering, Shear, and CMB lensing Analyses
BibTeX
@misc{mandal2026occams,
title = {{OCCAMS: A Multi-Task Learning-Based Investigation of the CMB Power Spectra}},
author = {Ankit Mandal and Rahul Shah and Md. Wali Hossain and Somasri Sen and Anjan A Sen and Supratik Pal},
howpublished = {arXiv},
year = {2026},
month = oct,
doi = {10.48550/arxiv.2610.40415},
eprint = {2610.40415},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2610.40415},
}
RIS
TY - UNPB
TI - OCCAMS: A Multi-Task Learning-Based Investigation of the CMB Power Spectra
AU - Ankit Mandal
AU - Rahul Shah
AU - Md. Wali Hossain
AU - Somasri Sen
AU - Anjan A Sen
AU - Supratik Pal
PY - 2026
DA - 2026/10/29
PB - arXiv
DO - 10.48550/arxiv.2610.40415
UR - https://arxiv.org/abs/2610.40415
AB - How many latent variables are needed to fit the cosmic microwave background (CMB) temperature and polarisation power spectra when a neural network is given no cosmological model, and to what extent do they map onto the standard cosmological parameters? We use multi-task learning with variational autoencoders trained on CLASS-generated spectra (TT, TE, EE) over 2≤ℓ≤2500. No cosmological model is built into the network, and the number of latent variables, L, is varied over L=5, $6$ and $7$. Their fits are compared to the Planck 2018 data using χ̂^2, AIC, BIC and Bayesian evidence. Their correspondence with cosmological parameters is measured with normalised mutual information post-training. The loss weights are set within a single training run by uncertainty weighting, with the weight of the disentanglement term either fixed (semi-automated, SA) or learned (completely automated, CA). The reconstruction loss is split over multipole ranges in three configurations that separate the roles of temperature and polarisation and of the low- and high-ℓ ranges. In four of the six combinations of configuration and strategy, the fit improves from L=5 to L=6 and not beyond, and L=6 is the best, or close to the best, in five of the six cases. So, six latent variables are needed, as many as the free parameters of the standard, model-dependent approach. SA gives clearer disentanglement, while CA needs no a priori choice of the disentanglement weight and has the higher evidence at L=5 and $6$: the two strategies are points on a trade-off between interpretability and fit to the data. Notably, the latent variables for τ and A_s share almost no information in any model, consistent with large-scale polarisation separating the two, as in standard analyses. A deep-learning approach can thus give a compact, data-driven parametrisation of the CMB spectra.
ER -