In the era of exploding survey volumes, traditional methods of spectroscopic analysis are being pushed to their limits. In response, we develop deep-REMAP, a novel deep learning framework that utilizes a regularized, multi-task approach to predict stellar atmospheric parameters from observed spectra. We train a deep convolutional neural network on the PHOENIX synthetic spectral library and use transfer learning to fine-tune the model on a small subset of observed FGK dwarf spectra from the MARVELS survey. We then apply the model to 732 uncharacterized FGK giant candidates from the same survey. When validated on 30 MARVELS calibration stars, deep-REMAP accurately recovers the effective temperature ($T_{\rm{eff}}$), surface gravity ($\log \rm{g}$), and metallicity ([Fe/H]), achieving a precision of, for instance, approximately 75 K in $T_{\rm{eff}}$. By combining an asymmetric loss function with an embedding loss, our regression-as-classification framework is interpretable, robust to parameter imbalances, and capable of capturing non-Gaussian uncertainties. While developed for MARVELS, the deep-REMAP framework is extensible to other surveys and synthetic libraries, demonstrating a powerful and automated pathway for stellar characterization.

Foundation models for structured time series data must contend with a fundamental challenge: observations often conflate the true underlying physical phenomena with systematic distortions introduced by measurement instruments. This entanglement limits model generalization, especially in heterogeneous or multi-instrument settings. We present a causally-motivated foundation model that explicitly disentangles physical and instrumental factors using a dual-encoder architecture trained with structured contrastive learning. Leveraging naturally occurring observational triplets (i.e., where the same target is measured under varying conditions, and distinct targets are measured under shared conditions) our model learns separate latent representations for the underlying physical signal and instrument effects. Evaluated on simulated astronomical time series designed to resemble the complexity of variable stars observed by missions like NASA's Transiting Exoplanet Survey Satellite (TESS), our method significantly outperforms traditional single-latent space foundation models on downstream prediction tasks, particularly in low-data regimes. These results demonstrate that our model supports key capabilities of foundation models, including few-shot generalization and efficient adaptation, and highlight the importance of encoding causal structure into representation learning for structured data.

Context: Astronomy and astrophysics demand rigorous handling of uncertainties to ensure the credibility of outcomes. The growing integration of artificial intelligence offers a novel avenue to address this necessity. This convergence presents an opportunity to create advanced models capable of quantifying diverse sources of uncertainty and automating complex data relationship exploration. What: We introduce a hierarchical Bayesian architecture whose probabilistic relationships are modeled by neural networks, designed to forecast stellar attributes such as mass, radius, and age (our main target). This architecture handles both observational uncertainties stemming from measurements and epistemic uncertainties inherent in the predictive model itself. As a result, our system generates distributions that encapsulate the potential range of values for our predictions, providing a comprehensive understanding of their variability and robustness. Methods: Our focus is on dating main sequence stars using a technique known as Chemical Clocks, which serves as both our primary astronomical challenge and a model prototype. In this work, we use hierarchical architectures to account for correlations between stellar parameters and optimize information extraction from our dataset. We also employ Bayesian neural networks for their versatility and flexibility in capturing complex data relationships. Results: By integrating our machine learning algorithm into a Bayesian framework, we have successfully propagated errors consistently and managed uncertainty treatment effectively, resulting in predictions characterized by broader uncertainty margins. This approach facilitates more conservative estimates in stellar dating. Our architecture achieves age predictions with a mean absolute error of less than 1 Ga for the stars in the test dataset.

A method based on Generative Adversaria! Networks (GANs) is developed for disentangling the physical (effective temperature and gravity) and chemical (metallicity, overabundance of a-elements with respect to iron) atmospheric properties in astronomical spectra. Using a projection of the stellar spectra, commonly called latent space, in which the contribution dueto one or several main stellar physicochemical properties is minimised while others are enhanced, it was possible to maximise the information related to certain properties, which can then be extracted using artificial neural networks (ANN) as regressors with higher accuracy than a reference method based on the use of ANN trained with the original spectra. Methods. Our model utilises autoencoders, comprising two artificial neural networks: an encoder anda decoder which transform input data into a low-dimensional representation known as latent space. It also uses discriminators, which are additional neural networks aimed at transforming the traditional autoencoder training into an adversaria! approach, to disentangle or reinforce the astrophysical parameters from the latent space. The GANDALF tool is described. It was developed to define, train, and test our GAN model with a web framework to show how the disentangling algorithm works visually. It is open to the community in Github. Results. The performance of our approach for retrieving atmospheric stellar properties from spectra is demonstrated using Gaia Radial Velocity Spectrograph (RVS) data from DR3. We use a data-driven perspective and obtain very competitive values, ali within the literature errors, and with the advantage of an important dimensionality reduction of the data to be processed.

Deep learning (DL) techniques are a promising approach among the set of methods used in the ever-challenging determination of stellar parameters in M dwarfs. In this context, transfer learning could play an important role in mitigating uncertainties in the results due to the synthetic gap (i.e. difference in feature distributions between observed and synthetic data). We propose a feature-based deep transfer learning (DTL) approach based on autoencoders to determine stellar parameters from high-resolution spectra. Using this methodology, we provide new estimations for the effective temperature, surface gravity, metallicity, and projected rotational velocity for 286 M dwarfs observed by the CARMENES survey. Using autoencoder architectures, we projected synthetic PHOENIX-ACES spectra and observed CARMENES spectra onto a new feature space of lower dimensionality in which the differences between the two domains are reduced. We used this low-dimensional new feature space as input for a convolutional neural network to obtain the stellar parameter determinations. We performed an extensive analysis of our estimated stellar parameters, ranging from 3050 to 4300 K, 4.7 to 5.1 dex, and -0.53 to 0.25 dex for Teff, logg, and [Fe/H], respectively. Our results are broadly consistent with those of recent studies using CARMENES data, with a systematic deviation in our Teff scale towards hotter values for estimations above 3750 K. Furthermore, our methodology mitigates the deviations in metallicity found in previous DL techniques due to the synthetic gap. We consolidated a DTL-based methodology to determine stellar parameters in M dwarfs from synthetic spectra, with no need for high-quality measurements involved in the knowledge transfer. These results suggest the great potential of DTL to mitigate the differences in feature distributions between the observations and the PHOENIX-ACES spectra.

The properties of exoplanet host stars are traditionally characterized through a detailed forward-modeling analysis of high-resolution spectra. However, many exoplanet radial velocity surveys employ iodine-cell-calibrated spectrographs, such that the vast majority of spectra obtained include an imprinted forest of iodine absorption lines. For surveys that use iodine cells, iodine-free "template" spectra must be separately obtained for precise stellar characterization. These template spectra often require extensive additional observing time to obtain, and they are not always feasible to obtain for faint stars. In this paper, we demonstrate that machine learning methods can be applied to infer stellar parameters and chemical abundances from iodine-imprinted spectra with high accuracy and precision. The methods presented in this work are broadly applicable to any iodine-cell-calibrated spectrograph. We make publicly available our spectroscopic pipeline, the Cannon HIRES Iodine Pipeline (CHIP), which derives stellar parameters and 15 chemical abundances from iodine-imprinted spectra of FGK stars and which has been set up for ease of use with Keck/HIRES spectra. Our proof-of-concept offers an efficient new avenue to rapidly estimate a large number of stellar parameters even in the absence of an iodine-free template spectrum.

Traditional spectral analysis methods are increasingly challenged by the exploding volumes of data produced by contemporary astronomical surveys. In response, we develop deep-Regularized Ensemble-based Multi-task Learning with Asymmetric Loss for Probabilistic Inference ($\rm{deep-REMAP}$), a novel framework that utilizes the rich synthetic spectra from the PHOENIX library and observational data from the MARVELS survey to accurately predict stellar atmospheric parameters. By harnessing advanced machine learning techniques, including multi-task learning and an innovative asymmetric loss function, $\rm{deep-REMAP}$ demonstrates superior predictive capabilities in determining effective temperature, surface gravity, and metallicity from observed spectra. Our results reveal the framework's effectiveness in extending to other stellar libraries and properties, paving the way for more sophisticated and automated techniques in stellar characterization.

The Gaia Data Release 3 (DR3), published in June 2022, delivers a diverse set of astrometric, photometric, and spectroscopic measurements for more than a billion stars. The wealth and complexity of the data makes traditional approaches for estimating stellar parameters for the full Gaia dataset almost prohibitive. We have explored different supervised learning methods for extracting basic stellar parameters as well as distances and line-of-sight extinctions, given spectro-photo-astrometric data (including also the new Gaia XP spectra). For training we use an enhanced high-quality dataset compiled from Gaia DR3 and ground-based spectroscopic survey data covering the whole sky and all Galactic components. We show that even with a simple neural-network architecture or tree-based algorithm (and in the absence of Gaia XP spectra), we succeed in predicting competitive results (compared to Bayesian isochrone fitting) down to faint magnitudes. We will present a new Gaia DR3 stellar-parameter catalogue obtained using the currently best-performing machine-learning algorithm for tabular data, XGBoost, in the near future.

We discuss how to achieve mapping from large sets of imperfect simulations and observational data with unsupervised domain adaptation. Under the hypothesis that simulated and observed data distributions share a common underlying representation, we show how it is possible to transfer between simulated and observed domains. Driven by an application to interpret stellar spectroscopic sky surveys, we construct the domain transfer pipeline from two adversarial autoencoders on each domains with a disentangling latent space, and a cycle-consistency constraint. We then construct a differentiable pipeline from physical stellar parameters to realistic observed spectra, aided by a supplementary generative surrogate physics emulator network. We further exemplify the potential of the method on the reconstructed spectra quality and to discover new spectral features associated to elemental abundances.

Using a machine learning technique, a team of Instituto de Astrofísica e Ciências do Espaço researchers constrained the radius of an exoplanet with known mass. Solène Ulmer-Moll, a Ph.D. student at the Science Faculty of the University of Porto (FCUP), explains this result was obtained by using knowledge from different fields: "This novel way to forecast exoplanet radius is a perfect example of the synergy between exoplanet science and machine learning techniques." To characterize a planet, both its mass and radius are needed in order to find the planet's density, and from that, to infer its composition. But both data are only available for a reduced number of exoplanets, since the mass is often determined by radial velocity measurements, while radius is measured with the transit method. The team developed an algorithm that accurately forecasts the radius of a wide range of exoplanets, if several other planetary and stellar parameters are known, including the exoplanet's mass and equilibrium temperature.