Astronomical image deconvolution has evolved from early linear regularization to modern deep learning approaches. This article provides a concise overview of the historical milestones, methodological breakthroughs, and current challenges in the field, with emphasis on sparsity, compressed sensing, and the role of deep learning in large astronomical surveys.

Weak gravitational lensing of background galaxies traces the growth of dark matter and is key to constraining the cosmological model. This proceedings paper provides an overview of the analysis presented in Voigt (2024), applying convolutional neural networks (CNNs) to the challenge of precision shear estimation. Using a shallow CNN, we investigate the sensitivity of shear recovery to point spread function (PSF) misestimation, with reference to the accuracy requirements of Euclid. To mitigate noise bias we employ a non-standard loss function, and assess performance assuming a linear relation between predicted and true shears.

Photometric redshift estimation (photo-z) offers a scalable alternative to spectroscopic measurements, at the expense of reduced accuracy and increased uncertainty. In this work, we investigate machine learning approaches that explicitly incorporate uncertainty quantification into photo-z estimation using Sloan Digital Sky Survey (SDSS) data. Starting from a baseline neural network, we extend to Gaussian regression, quantile regression, and conformalized quantile regression (CQR). Our experiments demonstrate that, while the baseline model achieves competitive accuracy, uncertainty-aware methods provide more reliable and interpretable predictions.

The intracluster medium (ICM) records the history of galaxy clusters through its complex dynamical properties. To effectively interpret these properties, robust methods are needed to compare observational data with theoretical models. We present a novel machine learning framework for comparing ICM line-of-sight velocity maps derived from X-ray observations. Our approach uses convolutional and Siamese neural networks to identify similarities between different kinematic fields. We outline the architecture of this framework and perform a series of sanity checks to validate its performance. These checks demonstrate the model’s ability to correctly identify and quantify kinematic features, establishing a powerful new tool for future comparative studies of the ICM.

One of the significant problems in cosmology is the morphological classification of galaxies in order to gain insights about the origin and evolution of our Cosmos. Here, we investigate the classification accuracy of several modern convolutional neural network architectures (some of them pretrained on non-astronomical data and others trained from scratch to galaxy data), and of transformers with primary goal to understand their performance on conditions of fast training (i.e., in a few rounds). We experimented with Transfer Learning, and Vision Transformers (ViT). We contrasted their performance with that of traditional Convolutional Neural Networks (CNN) methods, i.e., AlexNet, and VGG. Our results show a 75% classification accuracy (by the ViT and VGG), beating the rest of the competitors. Moreover, one pretrained VGG model showed also excellent performance.

Efforts are underway to measure the global 21 cm signal from neutral hydrogen, which is a powerful probe of the early universe, using NASA radio telescopes on the far side of the Moon. Physics-based models of the signal are computationally expensive to perform Bayesian multi-parameter inferences, for which we have developed novel, publicly-available neural network emulators utilizing a Long Short-Term Memory (LSTM) network and a Kolmogorov-Arnold Network (KAN). <sc>21cmLSTM</sc> is currently the most accurate emulator in the community by leveraging the signal’s temporally-correlated structure, and <sc>21cmKAN</sc> maintains similar accuracy while training 75 times faster, by learning expressive functional transformations. Each emulator can fit realistic mock signals and obtain unbiased physical parameter constraints, with <sc>21cmKAN</sc> able to complete end-to-end training and inference in under 30 minutes. The implementation of machine learning tools like these in data analysis pipelines is important to fully exploit upcoming measurements of the cosmological 21 cm signal.

We present ProMage, a feed-forward neural network that emulates the computation of observer- and rest-frame magnitudes from the SED generation package ProSpect. The network predicts magnitudes conditioned on input galaxy physical properties, including redshift, star formation history, gas and dust parameters. ProMage accelerates magnitude computation by a factor of $10^4$ compared to ProSpect, while achieving per-mille relative accuracy for 99% of sources in the test set across the g,r,i,z,y Hyper Suprime-Cam bands. This acceleration is key to enabling fast inference of galaxy physical properties in next-generation Stage IV surveys and to generating large catalogue realisations in forward-modelling frameworks such as GalSBI-SPS.

We present a deep learning-based framework for automated quasar spectral analysis, addressing two key challenges: continuum estimation and absorption feature detection. Using autoencoders, convolutional neural networks (CNNs), and U-Net architectures, we demonstrate improved performance in reconstructing the quasar continuum and identifying absorption lines. These models are trained to learn complex, non-linear correlations across wide spectral regions, enabling robust predictions even in the presence of noise or missing data. Our methods offer greater flexibility and accuracy, particularly in low signal-to-noise regimes.

The U-Net architecture, in particular, shows strong potential for precise absorption feature detection, leveraging its encoder-decoder structure with skip connections to capture both local and global spectral patterns. This enables the identification of absorption signatures associated with metal-enriched gas in the circumgalactic medium. Our results highlight the scalability of these techniques for application to large-scale surveys such as DESI and WEAVE, making progress for new insights into the distribution and dynamics of baryons in the Universe.

Information field theory (IFT) is the application of probabilistic reasoning to fields. Physical fields are mathematical functions over continuous spaces that exhibit certain properties of regularity, such as limited variance and finite gradients. Inferring a field from an observational dataset should exploit these regularities. However, the finite number of constraints that the data provides is insufficient to determine the infinite number of degrees of freedom of a field. IFT enables us to derive optimal field inference algorithms that explicitly exploit domain knowledge. These algorithms can be implemented via Numerical Information Field Theory (NIFTy). In NIFTy, neural operator forward models can be written and inverted probabilistically. NIFTy thereby infers fields and their remaining uncertainties. This is achieved using novel variational inference schemes that scale quasi-linearly, even for ultra-high dimensional problems. This paper introduces the basic concepts of IFT and NIFTy, highlights a few of their astrophysical applications, and discusses their artificial intelligence (AI) perspective. Finally, UBIK (the Universal Bayesian Imaging Kit), an emerging customisation of NIFTy for a suite of astrophysical telescopes, is presented as a central tool to the topic of the UniversAI conference.

Machine learning (ML) is increasingly important in astrophysics, enabling the analysis of large, complex datasets for tasks such as stellar classification and gravitational wave detection. Yet many ML models are highly non-linear, with intricate feature interactions that make them difficult to interpret and often viewed as “black boxes.” Interpretability is therefore essential for ensuring trust, scientific validity, but also for uncovering new scientific insights.

Global feature effect methods address this by quantifying how input features influence model outputs on average. However, such explanations may be misleading when strong interactions cause individual behaviours to deviate from the average. Regional effect plots mitigate this by partitioning the feature space into more homogeneous subspaces, producing more accurate effect estimates.

This talk reviews global feature effect methods, introduces recent advances, and presents Effector, an open-source Python package for efficiently generating accurate global and regional feature effect plots.

This extended abstract, structured as a series of ten questions, briefly summarizes the content of my keynote talk, which aimed to introduce astrophysicists to the area of ontology-based data access, a promising paradigm for designing intelligent information systems which has been extensively studied within the symbolic AI and database communities over past twenty years.

We present a new framework for multi-line analysis that combines kernel principal component analysis (kernel PCA) and kernel shapley additive explanations (kernel SHAP). Applied to ALMA Band 3 data of NGC 1068, kernel PCA reveals features—including the AGN-driven outflow —not clearly captured by PCA, while kernel SHAP quantifies each line’s contribution. These results are consistent with previous studies and confirmed by a two-sample t-test. Moreover, kernel PCA scores correlate with LTE-derived column densities, highlighting the robustness of this framework.

Given the potential of Large Language Models (LLMs) to transform how people engage with information systems across various industries, we have explored ways to enhance functionality within the context of astronomy science archives. Searching for specific types of object within archival data is a common request from archive users. However, this feature is typically unavailable due to the challenges of classifying the billions of objects present in astronomical catalogues. To address this need from astronomers, we present a proof-of-concept designed to potentially improve search functionalities within the ESA Astronomy Science Archives. The prototype makes use of the new EVA (ESA Virtual Assistant) LLM engine powered by ESA Datalabs, allowing EVA bots to experiment with advanced LLM and Agentic RAG (Retrieval Augmented Generation) capabilities. By utilizing data from the CDS SIMBAD astronomical object database, the proposed solution demonstrates how our implementation could offer astronomers enhanced functionalities when using the EVA bot in ESASky.

This paper presents an innovative approach to teaching Artificial Intelligence (AI) thinking to primary students through hands-on activities using spreadsheets, developed as part of the Stardusts Science Education Program. The program aims to demystify core AI concepts—prediction, classification, and learning—by demonstrating how AI works rather than just how to use it. Utilizing familiar tools like spreadsheets and real-world analogies, the curriculum is structured into four levels: Classification Basics, Recommendation Systems, Spreadsheet Neural Networks, and Galaxy Classification. Each level incorporates gamification and scaffolding to make abstract mathematical concepts tangible. Key activities include human classification games, group-based prediction for recommendation systems, building simple neural networks in spreadsheets to predict crop yield, and applying AI concepts to classify galaxy images. The program emphasizes connecting daily life examples (e.g., YouTube algorithms) to scientific applications (e.g., astronomy), fostering curiosity and critical thinking. This initiative underscores the potential of spreadsheets as an accessible educational tool for introducing complex AI principles to young learners.

Predicting interplanetary coronal mass ejections (ICMEs) remains challenging due to limited datasets, the complexity of some approaches based on machine learning and the continued reliance on manual event validation. We introduce SWAAT (Space Weather Automated Analysis Tool), a framework that integrates automated validation with synthetic data-driven prediction to improve ICME transit time (TT) forecasts based on propagation velocity (Vp). The system comprises two main modules: (i) The Automated Data Acquisition and Analysis (ADAA), which automates ICME candidate validation using GOES X-ray and SDO imagery, and (ii) The Generative-Evolutionary Synthetic Data (GESD), which employs evolutionary algorithms to generate optimized synthetic datasets for model training. By combining automatic event validation with quality synthetic data, SWAAT mitigates the limitations imposed by scarce datasets thanks to the contribution of synthetic data, improving the accuracy of space weather forecasting.

GANexo v2.0 extends our previous framework for exoplanet transit analysis by moving from synthetic signal generation to direct application on real observational data. The new version combines autoencoder-based noise reduction with Generative Adversarial Network (GAN) models, specifically a conditional GAN (cGAN) used for regression, to extract key transit parameters from noisy light curves. This integration improves robustness against observational noise, enhances parameter estimation accuracy, and provides a reliable AI-driven tool specially for ground-based exoplanet studies.

This position paper frames strategic space logistics as a game among autonomous agencies that may delegate portions of their campaigns and exchange incentives through bilateral agreements. Motivated by recent trends in space commercialization and campaign-level mission design, we introduce a formal noncooperative model in which each agency controls a workload and limited resources; agencies may outsource fractions of work to partners at a cost, and they allocate incentives to secure cooperation. Our aim is to motivate and orient theoretical and simulation research at the intersection of space logistics, game theory, and computational economics. For that, this paper advances a research agenda.

The Gaia mission has observed over 2 billion stars repeatedly across the entire sky over 10 years, revealing the many astronomical objects that vary on human timescales from seconds to years. Its repeated astrometric, photometric, spectrophotometric and spectroscopic measurements create an unprecedented dataset to probe the variable celestial sources down to $G\approx 21$ mag. To extract meaningful results from these many time series for so many sources, we have used machine learning techniques for crossmatching, variability detection, and variability classification. This approach has now led to the largest catalogue of classified variable sources ever produced over the entire celestial sphere.

Machine learning (ML) is revolutionising the analysis of large astronomical surveys such as Gaia DR3. We present a two-step ML pipeline to classify Asymptotic Giant Branch (AGB) stars: (i) cosine similarity identifies candidate AGB stars from 140 million sources, and (ii) an XGBoost classifier which separates carbon-rich (C-rich) and oxygen-rich (O-rich) stars. Our classifier achieves ROC-AUC > 99.9%, producing a catalogue of 307,191 O-rich and 4,227 C-rich stars. Feature importance analysis shows that the sixth RP coefficient (RP6) dominates classification, underlining the critical role of Gaia red spectral features. This work demonstrates how hybrid ML approaches can efficiently identify chemically distinct stellar populations in massive survey datasets, providing opportunities for new studies of Galactic structure and stellar evolution.