A key objective for upcoming surveys, and when re-analysing archival data, is the identification of variable stellar sources. However, the selection of these sources is often complicated by the unavailability of light curve data. Utilising a self-organising map (SOM), we demonstrate the selection of diverse variable source types from a catalogue of variable and non-variable SDSS Stripe 82 sources whilst employing only the median $u-g$, $g-r$, $r-i$, and $i-z$ photometric colours for each source as input, without using source magnitudes. This includes the separation of main sequence variable stars that are otherwise degenerate with non-variable sources ($u-g$,$g-r$) and ($r-i$,$i-z$) colour-spaces. We separate variable sources on the main sequence from all other variable and non-variable sources with a purity of $80.0\%$ and completeness of $25.1\%$, figures which can be modified depending on the application. We also explore the varying ability of the same method to simultaneously select other types of variable sources from the heterogeneous sample, including variable quasars and RR-Lyrae stars. The demonstrated ability of this method to select variable main sequence stars in colour-space holds promise for application in future survey reduction pipelines and for the analysis of archival data, where light curves may not be available or may be prohibitively expensive to obtain.

Since the turn of the century, astronomers have been exploiting the rich information afforded by combining stellar kinematic maps and imaging in an attempt to recover the intrinsic, three-dimensional (3D) shape of a galaxy. A common intrinsic shape recovery method relies on an expected monotonic relationship between the intrinsic misalignment of the kinematic and morphological axes and the triaxiality parameter. Recent studies have, however, cast doubt about underlying assumptions relating shape and intrinsic kinematic misalignment. In this work, we aim to recover the 3D shape of individual galaxies using their projected stellar kinematic and flux distributions using a supervised machine learning approach with mixture density network (MDN). Using a mock dataset of the EAGLE hydrodynamical cosmological simulation, we train the MDN model for a carefully selected set of common kinematic and photometric parameters. Compared to previous methods, we demonstrate potential improvements achieved with the MDN model to retrieve the 3D galaxy shape along with the uncertainties, especially for prolate and triaxial systems. We make specific recommendations for recovering galaxy intrinsic shapes relevant for current and future integral field spectroscopic galaxy surveys.

We present source detection and catalogue construction pipelines to build the first catalogue of radio galaxies from the 270 $\rm deg^2$ pilot survey of the Evolutionary Map of the Universe (EMU-PS) conducted with the Australian Square Kilometre Array Pathfinder (ASKAP) telescope. The detection pipeline uses Gal-DINO computer vision networks (Gupta et al. 2024, PASA, 41, e001) to predict the categories of radio morphology and bounding boxes for radio sources, as well as their potential infrared host positions. The Gal-DINO network is trained and evaluated on approximately 5 000 visually inspected radio galaxies and their infrared hosts, encompassing both compact and extended radio morphologies. We find that the Intersection over Union (IoU) for the predicted and ground-truth bounding boxes is larger than 0.5 for 99% of the radio sources, and 98% of predicted host positions are within $3^{\prime \prime}$ of the ground-truth infrared host in the evaluation set. The catalogue construction pipeline uses the predictions of the trained network on the radio and infrared image cutouts based on the catalogue of radio components identified using the Selavy source finder algorithm. Confidence scores of the predictions are then used to prioritise Selavy components with higher scores and incorporate them first into the catalogue. This results in identifications for a total of 211 625 radio sources, with 201 211 classified as compact and unresolved. The remaining 10 414 are categorised as extended radio morphologies, including 582 FR-I, 5 602 FR-II, 1 494 FR-x (uncertain whether FR-I or FR-II), 2 375 R (single-peak resolved) radio galaxies, and 361 with peculiar and other rare morphologies. Each source in the catalogue includes a confidence score. We cross-match the radio sources in the catalogue with the infrared and optical catalogues, finding infrared cross-matches for 73% and photometric redshifts for 36% of the radio galaxies. The EMU-PS catalogue and the detection pipelines presented here will be used towards constructing catalogues for the main EMU survey covering the full southern sky.

In large-scale galaxy surveys, particularly deep ground-based photometric studies, galaxy blending was inevitable. Such blending posed a potential primary systematic uncertainty for upcoming surveys. Current deblenders predominantly depended on analytical modelling of galaxy profiles, facing limitations due to inflexible and imprecise models. We presented a novel approach, using a U-net structured transformer-based network for deblending astronomical images, which we term the CAT-deblender. It was trained using both RGB and the grz-band images, spanning two distinct data formats present in the Dark Energy Camera Legacy Survey (DECaLS) database, including galaxies with diverse morphologies in the training dataset. Our method necessitated only the approximate central coordinates of each target galaxy, sourced from galaxy detection, bypassing assumptions on neighbouring source counts. Post-deblending, our RGB images retained a high signal-to-noise peak, consistently showing superior structural similarity against ground truth. For multi-band images, the ellipticity of central galaxies and median reconstruction error for r-band consistently lie within $\pm$0.025 to $\pm$0.25, revealing minimal pixel residuals. In our comparison of deblending capabilities focused on flux recovery, our model showed a mere 1% error in magnitude recovery for quadruply blended galaxies, significantly outperforming SExtractor’s higher error rate of 4.8%. Furthermore, by cross-matching with the publicly accessible overlapping galaxy catalogs from the DECaLS database, we successfully deblended 433 overlapping galaxies. Moreover, we have demonstrated effective deblending of 63 733 blended galaxy images, randomly chosen from the DECaLS database.

Creating radio galaxy catalogues from next-generation deep surveys requires automated identification of associated components of extended sources and their corresponding infrared hosts. In this paper, we introduce RadioGalaxyNET, a multimodal dataset, and a suite of novel computer vision algorithms designed to automate the detection and localization of multi-component extended radio galaxies and their corresponding infrared hosts. The dataset comprises 4 155 instances of galaxies in 2 800 images with both radio and infrared channels. Each instance provides information about the extended radio galaxy class, its corresponding bounding box encompassing all components, the pixel-level segmentation mask, and the keypoint position of its corresponding infrared host galaxy. RadioGalaxyNET is the first dataset to include images from the highly sensitive Australian Square Kilometre Array Pathfinder (ASKAP) radio telescope, corresponding infrared images, and instance-level annotations for galaxy detection. We benchmark several object detection algorithms on the dataset and propose a novel multimodal approach to simultaneously detect radio galaxies and the positions of infrared hosts.

The present work discusses the use of a weakly-supervised deep learning algorithm that reduces the cost of labelling pixel-level masks for complex radio galaxies with multiple components. The algorithm is trained on weak class-level labels of radio galaxies to get class activation maps (CAMs). The CAMs are further refined using an inter-pixel relations network (IRNet) to get instance segmentation masks over radio galaxies and the positions of their infrared hosts. We use data from the Australian Square Kilometre Array Pathfinder (ASKAP) telescope, specifically the Evolutionary Map of the Universe (EMU) Pilot Survey, which covered a sky area of 270 square degrees with an RMS sensitivity of 25–35 $\mu$Jy beam$^{-1}$. We demonstrate that weakly-supervised deep learning algorithms can achieve high accuracy in predicting pixel-level information, including masks for the extended radio emission encapsulating all galaxy components and the positions of the infrared host galaxies. We evaluate the performance of our method using mean Average Precision (mAP) across multiple classes at a standard intersection over union (IoU) threshold of 0.5. We show that the model achieves a mAP$_{50}$ of 67.5% and 76.8% for radio masks and infrared host positions, respectively. The network architecture can be found at the following link: https://github.com/Nikhel1/Gal-CAM

We present a comparison between the performance of a selection of source finders (SFs) using a new software tool called Hydra. The companion paper, Paper I, introduced the Hydra tool and demonstrated its performance using simulated data. Here we apply Hydra to assess the performance of different source finders by analysing real observational data taken from the Evolutionary Map of the Universe (EMU) Pilot Survey. EMU is a wide-field radio continuum survey whose primary goal is to make a deep ($20\mu$Jy/beam RMS noise), intermediate angular resolution ($15^{\prime\prime}$), 1 GHz survey of the entire sky south of $+30^{\circ}$ declination, and expecting to detect and catalogue up to 40 million sources. With the main EMU survey it is highly desirable to understand the performance of radio image SF software and to identify an approach that optimises source detection capabilities. Hydra has been developed to refine this process, as well as to deliver a range of metrics and source finding data products from multiple SFs. We present the performance of the five SFs tested here in terms of their completeness and reliability statistics, their flux density and source size measurements, and an exploration of case studies to highlight finder-specific limitations.

The latest generation of radio surveys are now producing sky survey images containing many millions of radio sources. In this context it is highly desirable to understand the performance of radio image source finder (SF) software and to identify an approach that optimises source detection capabilities. We have created Hydra to be an extensible multi-SF and cataloguing tool that can be used to compare and evaluate different SFs. Hydra, which currently includes the SFs Aegean, Caesar, ProFound, PyBDSF, and Selavy, provides for the addition of new SFs through containerisation and configuration files. The SF input RMS noise and island parameters are optimised to a 90% ‘percentage real detections’ threshold (calculated from the difference between detections in the real and inverted images), to enable comparison between SFs. Hydra provides completeness and reliability diagnostics through observed-deep ($\mathcal{D}$) and generated-shallow ($\mathcal{S}$) images, as well as other statistics. In addition, it has a visual inspection tool for comparing residual images through various selection filters, such as S/N bins in completeness or reliability. The tool allows the user to easily compare and evaluate different SFs in order to choose their desired SF, or a combination thereof. This paper is part one of a two part series. In this paper we introduce the Hydra software suite and validate its $\mathcal{D/S}$ metrics using simulated data. The companion paper demonstrates the utility of Hydra by comparing the performance of SFs using both simulated and real images.

We demonstrate that the Bayesian evidence can be used to find a good approximation of the ground truth likelihood function of a dataset, a goal of the likelihood-free inference (LFI) paradigm. As a concrete example, we use forward modelled sky-averaged 21-cm signal antenna temperature datasets where we artificially inject noise structures of various physically motivated forms. We find that the Gaussian likelihood performs poorly when the noise distribution deviates from the Gaussian case, for example, heteroscedastic radiometric or heavy-tailed noise. For these non-Gaussian noise structures, we show that the generalised normal likelihood is on a similar Bayesian evidence scale with comparable sky-averaged 21-cm signal recovery as the ground truth likelihood function of our injected noise. We therefore propose the generalised normal likelihood function as a good approximation of the true likelihood function if the noise structure is a priori unknown.

The amount and complexity of data delivered by modern galaxy surveys has been steadily increasing over the past years. New facilities will soon provide imaging and spectra of hundreds of millions of galaxies. Extracting coherent scientific information from these large and multi-modal data sets remains an open issue for the community and data-driven approaches such as deep learning have rapidly emerged as a potentially powerful solution to some long lasting challenges. This enthusiasm is reflected in an unprecedented exponential growth of publications using neural networks, which have gone from a handful of works in 2015 to an average of one paper per week in 2021 in the area of galaxy surveys. Half a decade after the first published work in astronomy mentioning deep learning, and shortly before new big data sets such as Euclid and LSST start becoming available, we believe it is timely to review what has been the real impact of this new technology in the field and its potential to solve key challenges raised by the size and complexity of the new datasets. The purpose of this review is thus two-fold. We first aim at summarising, in a common document, the main applications of deep learning for galaxy surveys that have emerged so far. We then extract the major achievements and lessons learned and highlight key open questions and limitations, which in our opinion, will require particular attention in the coming years. Overall, state-of-the-art deep learning methods are rapidly adopted by the astronomical community, reflecting a democratisation of these methods. This review shows that the majority of works using deep learning up to date are oriented to computer vision tasks (e.g. classification, segmentation). This is also the domain of application where deep learning has brought the most important breakthroughs so far. However, we also report that the applications are becoming more diverse and deep learning is used for estimating galaxy properties, identifying outliers or constraining the cosmological model. Most of these works remain at the exploratory level though which could partially explain the limited impact in terms of citations. Some common challenges will most likely need to be addressed before moving to the next phase of massive deployment of deep learning in the processing of future surveys; for example, uncertainty quantification, interpretability, data labelling and domain shift issues from training with simulations, which constitutes a common practice in astronomy.

We discuss the impact of Gaia, the cornerstone mission of the European Space Agency (ESA), on the calibration of the period–luminosity and luminosity–metallicity relations of Cepheids and RR Lyrae stars, with specific reference to data published as part of the most recent Gaia releases: Early Data Release 3 (EDR3), on 19 December 2020, and Data Release 3 (DR3) on 13 June 2022. We provide future perspectives for the Gaia mission, including extensions approved by ESA and a tentative schedule of the data releases that will take place in the next few years. We briefly present plans for cross-Coordination Unit processing of Gaia data of Cepheids and RR Lyrae stars for DR4 and conclude by outlining the expected improvement in astrometry at the end of the extended Gaia mission, which will help to further strengthen the calibration of the first rung of the cosmic distance ladder.

Forthcoming data from the Vera Rubin Observatory, Euclid and Roman telescopes are expected to increase the number of strong lenses by two orders of magnitude. With current discovery methods these would be accompanied by an even greater number of false positives. In that context we find that using an ensemble of classifiers would provide a more complete sample of high-purity lenses and present methods to post-process the outputs of such classifiers to give reliable probabilities that a given image contains a lens.

Current searches for galaxy-scale strong lenses focus on massive Luminous Red Galaxies but tend to overlook late-type lenses, in part because of their smaller Einstein radii. We take advantage of the superb seeing of the UNIONS survey in the r-band to perform an imaging search for edge-on late-type lenses. We use Convolutional Neural Networks trained with simulated observations composed of images of real galaxies from UNIONS and real sources from HST. Using 3600 square degrees of the survey we test ∼7 million galaxies and find 56 systems with obvious signs of lensing. In addition, we empirically estimate the true prevalence of lenses in UNIONS by visually inspecting 120,000 randomly chosen images in the survey. We find that the number of edge-on lenses we discover with CNNs is compatible with these estimates.

Galaxy-galaxy strong lensing (SL) systems provide a unique opportunity to test modified gravity theories. Deviations from General Relativity are encoded in the post-Newtonian parameter (γ). As a preparation for the upcoming data from the Vera Rubin Observatory Legacy Survey of Space and Time (LSST), our research group collected imaging data of SL systems from ground-based telescopes and conducted spectroscopic observations of 21 systems on the Southern Astrophysical Research (SOAR) Telescope to measure the lens velocity dispersions, σv. We briefly describe the semi-automated SL modelling of the systems in this sample and combine the results with σv from SOAR to derive an estimate for γ. Our preliminary results yield a value of $$\gamma= 1.17_{ - 0.33}^{ + 0.29}$$, which is consistent with General Relativity. Although the error bars are limited by the sample size, this result represents the first constraint on modified gravity obtained purely from ground-based data, with a sample completely independent from previous studies, and which allows for a self consistent end-to-end analysis.

The shape of emission lines in the optical spectra of star-forming galaxies reveals the kinematics of the diffuse gaseous component. We analyse the shape of prominent emission lines in a sample of $\sim$ 53000 star-forming galaxies from the Sloan Digital Sky Survey, focusing on departures from gaussianity. Departures from a single gaussian profile allow us to probe the motion of gas and to assess the role of outflows. The sample is divided into groups according to their stellar velocity dispersion and star formation rate (SFR). The spectra within each group are stacked to improve the signal-to-noise ratio of the emission lines, to remove individual signatures, and to enhance the effect of SFR on the shapes of the emission lines. The moments of the emission lines, including kurtosis and skewness, are determined. We find that most of the emission lines in strong star-forming systems unequivocally feature negative kurtosis. This signature is present in $\mathrm{H}\unicode{x03B2}$ , $\mathrm{H}\unicode{x03B1}$ , [N ii], and [S ii] in massive galaxies with high SFRs. We attribute it as evidence of radial outflows of ionised gas driven by the star formation of the galaxies. Also, most of the emission lines in low-mass systems with high SFRs feature negative skewness, and we interpret it as evidence of dust obscuration in the galactic disk. These signatures are however absent in the [O iii] line, which is believed to trace a different gas component. The observed trend is significantly stronger in face-on galaxies, indicating that star formation drives the outflows along the galactic rotation axis, presumably the path of least resistance. The data suggest that outflows driven by star formation exert accumulated impacts on the interstellar medium, and the outflow signature is more evident in older galaxies as they have experienced a longer total duration of star formation.

We present a set of peculiar radio sources detected using an unsupervised machine learning method. We use data from the Australian Square Kilometre Array Pathfinder (ASKAP) telescope to train a self-organizing map (SOM). The radio maps from three ASKAP surveys, Evolutionary Map of Universe pilot survey (EMU-PS), Deep Investigation of Neutral Gas Origins pilot survey (DINGO), and Survey With ASKAP of GAMA-09 + X-ray (SWAG-X), are used to search for the rarest or unknown radio morphologies. We use an extension of the SOM algorithm that implements rotation and flipping invariance on astronomical sources. The SOM is trained using the images of all ‘complex’ radio sources in the EMU-PS which we define as all sources catalogued as ‘multi-component’. The trained SOM is then used to estimate a similarity score for complex sources in all surveys. We select 0.5% of the sources that are most complex according to the similarity metric and visually examine them to find the rarest radio morphologies. Among these, we find two new odd radio circle (ORC) candidates and five other peculiar morphologies. We discuss multiwavelength properties and the optical/infrared counterparts of selected peculiar sources. In addition, we present examples of conventional radio morphologies including: diffuse emission from galaxy clusters, and resolved, bent-tailed, and FR-I and FR-II type radio galaxies. We discuss the overdense environment that may be the reason behind the circular shape of ORC candidates.

We demonstrate the effectiveness of a Bayesian evidence -based analysis for diagnosing and disentangling the sky-averaged 21-cm signal from instrumental systematic effects. As a case study, we consider a simulated REACH pipeline with an injected systematic. We demonstrate that very poor performance or erroneous signal recovery is achieved if the systematic remains unmodelled. These effects include sky-averaged 21-cm posterior estimates resembling a very deep or wide signal. However, when including parameterised models of the systematic, the signal recovery is dramatically improved in performance. Most importantly, a Bayesian evidence-based model comparison is capable of determining whether or not such a systematic model is needed as the true underlying generative model of an experimental dataset is in principle unknown. We, therefore, advocate a pipeline capable of testing a variety of potential systematic errors with the Bayesian evidence acting as the mechanism for detecting their presence.

A pilot project has been proceeded to map $1\, \textrm{deg}^2$ on the Galactic plane for radio recombination lines (RRLs) using the Five-hundred-metre Aperture Spherical Telescope (FAST). The motivation is to verify the techniques and reliabilities for a large-scale Galactic plane RRL survey with FAST aiming to investigate the ionised environment in the Galaxy. The data shows that the bandpass of the FAST 19 beam L-band is severely affected by radio frequency interferences and standing wave ripples, which can hardly be corrected by traditional low order polynomials. In this paper, we investigate a series of penalised least square (PLS) based baseline correction methods for radio astronomical spectra that usually contain weak signals with high level of noise. Three promising penalised least squares based methods, AsLS, arPLS, and asPLS are evaluated. Adopting their advantages, a modified method named rrlPLS is developed to optimise the baseline fitting to our RRL spectra. To check their effectiveness, the four methods are tested by simulations and further verified using observed data sets. It turns out that the rrlPLS method, with optimised parameter $\lambda=2\times10^8$ , reveals the most sensitive and reliable emission features in the RRL map. By injecting artificial line profiles into the real data cube, a further evaluation of profile distortion is conducted for rrlPLS. Comparing to simulated signals, the processed lines with low signal-to-noise ratio are less affected, of which the uncertainties are mainly caused by the rms noise. The rrlPLS method will be applied for baseline correction in future data processing pipeline of FAST RRL survey. Configured with proper parameters, the rrlPLS technique verified in this work may also be used for other spectroscopy projects.

Understanding the physical and evolutionary properties of Hot Stellar Systems (HSS) is a major challenge in astronomy. We studied the dataset on 13 456 HSS of Misgeld & Hilker (2011, MNRAS, 414, 3 699) that includes 12 763 candidate globular clusters using stellar mass ( $M_s$ ), effective radius ( $R_e$ ) and mass-to-luminosity ratio ( $M_s/L_\nu$ ), and found multi-layered homogeneous grouping among these stellar systems. Our methods elicited eight homogeneous ellipsoidal groups at the finest sub-group level. Some of these groups have high overlap and were merged through a multi-phased syncytial algorithm motivated from Almodóvar-Rivera & Maitra (2020, JMLR, 21, 1). Five groups were merged in the first phase, resulting in three complex-structured groups. Our algorithm determined further complex structure and permitted another merging phase, revealing two complex-structured groups at the highest level. A nonparametric bootstrap procedure was also used to estimate the confidence of each of our group assignments. These assignments generally had high confidence in classification, indicating great degree of certainty of the HSS assignments into our complex-structured groups. The physical and kinematic properties of the two groups were assessed in terms of $M_s$ , $R_e$ , surface density and $M_s/L_\nu$ . The first group consisted of older, smaller and less bright HSS while the second group consisted of brighter and younger HSS. Our analysis provides novel insight into the physical and evolutionary properties of HSS and also helps understand physical and evolutionary properties of candidate globular clusters. Further, the candidate globular clusters (GCs) are seen to have very high chance of really being GCs rather than dwarfs or dwarf ellipticals that are also indicated to be quite distinct from each other.

In this review, we introduce our recent applications of deep learning to solar and space weather data. We have successfully applied novel deep learning methods to the following applications: (1) generation of solar farside/backside magnetograms and global field extrapolation based on them, (2) generation of solar UV/EUV images from other UV/EUV images and magnetograms, (3) denoising solar magnetograms using supervised learning, (4) generation of UV/EUV images and magnetograms from Galileo sunspot drawings, (5) improvement of global IRI TEC maps using IGS TEC ones, (6) one-day forecasting of global TEC maps through image translation, (7) generation of high-resolution magnetograms from Ca II K images, (8) super-resolution of solar magnetograms, (9) flare classification by CNN and visual explanation by attribution methods, and (10) forecasting GOES solar X-ray profiles. We present major results and discuss them. We also present future plans for integrated space weather models based on deep learning.