Search arXiv⌕ Search
Home/Archive/arXiv

arXiv · 2502.18345

A Deep Neural Network Approach to Compact Source Removal

M. Madarász·G. Marton·I. Gezer·S. Lehner·J. Roquette·M. Audard·D. Hernandez·O. Dionatos
View original source ↗Read source document ↗Export metadata ↓

Abstract

Analysing extended emission in photometric observations of star-forming regions requires maps free from compact foreground, embedded, and background sources, which can interfere with various techniques used to characterise the interstellar medium. Within the framework of the NEMESIS project, we apply machine learning techniques to improve our understanding of the star formation timescales, which involves the unbiased analysis of the extended emission in these regions. We present a deep learning-based method for separating the signals of compact sources and extended emission in photometric observations made by the Herschel Space Observatory, facilitating the analysis of extended emission and improving the photometry of compact sources. Central to our approach is a modified U-Net architecture with partial convolutional layers. This method enables effective source removal and background estimation across various flux densities, using a series of partial convolutional layers, batch normalisation, and ReLU activation layers within blocks. Our training process utilised simulated sources injected into Herschel images, with controlled flux densities against known backgrounds. A dynamic, signal-to-noise ratio-based adaptive masking system was implemented to assess how prominently a compact source stands out from the surrounding background. The results demonstrate that our method can significantly improve the photometric accuracy in the presence of highly fluctuating backgrounds. Moreover, the approach can preserve all characteristics of the images, including the noise properties. The presented approach allows users to analyse extended emission without the interference of disturbing point sources or perform more precise photometry of sources located in complex environments. We also provide a Python tool with tutorials and examples to help the community effectively utilise this method.

Explore related subjects

astro-ph.IM

Keep this discovery

Explore connections, maps & timelines

BibTeXRIS

M. Madarász, G. Marton, I. Gezer, S. Lehner, J. Roquette, M. Audard, D. Hernandez, O. Dionatos. 2025-02-28. A Deep Neural Network Approach to Compact Source Removal. https://doi.org/10.1051/0004-6361%2F202453262

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related papers

arXiv2026

GW-YOLO: Multi-transient segmentation in LIGO using computer vision

Time series data and their time-frequency representations from gravitational-wave interferometers present opportunities for artificial intelligence methods in signal and image processing, particularly for low-latency analysis. In this work, we introduce GW-YOLO, a signal and noise identification tool based on the YOLO (You Only Look Once) object detection framework. GW-YOLO identifies whether an observed transient contains noise, an astrophysical signal, or both, while providing time-frequency coordinates of detected objects through pixel-level segmentation masks. Our approach achieves a 53% detection efficiency for binary black hole signals in the signal-to-noise ratio (SNR) 12--15 range when they overlap with transient noise, increasing to more than 75% at SNR 15--18. For binary neutron star signals overlapping with transient noise, the detection efficiency reaches 57% at SNR 30--33 and 87% at SNR 39--42. To our knowledge, this is the first quantitative assessment of the ability to detect astrophysical signals overlapping with realistic instrumental noise in gravitational-wave interferometers. We also present a fully automated, low-latency pipeline that produces pixel-level segmentation masks for individual noise transients and astrophysical signals, enabling further automation of event validation and downstream noise-mitigation procedures.

astro-ph.IM↗
arXiv2026

Multi-Scale Contrastive Attention for Light-Curve Representation Learning

Current and next-generation time-domain surveys demand automated techniques capable of analyzing millions of light curves, observed in multiple filters, without relying on exhaustive human annotation or scarce spectroscopic follow-up. We present Astra-CLR, an attention-based, self-supervised contrastive learning framework which enables the representation of raw light curves into a highly discriminative latent space. Pre-trained on $\sim$2.1 million unlabeled Zwicky Transient Facility light curves, the framework utilizes partial light curves as input sequences to generate asymmetric, multi-scale temporal views (explicitly contrasting shorter sequences against longer ones) forcing the network to learn a robust "local-to-global" mapping strategy. Furthermore, we introduce a novel multi-view late fusion architecture that extends the model to efficiently handle longer light curves with larger numbers of observations while accommodating the different cadences associated with each filter. The discriminatory power of the resulting representations was evaluated by using them as input to a Multinomial Logistic Regression classifier, trained to identify 12 broad classes of variability. Final accuracy achieved $\sim 0.70$. When applying a label-efficient, partial top-layer fine-tuning strategy, the topological structure of the latent space is significantly refined, boosting results to $\sim$0.77. Astra-CLR is the first publicly available multi-filter time-series Transformer trained exclusively on real ZTF light curves. Results presented here demonstrate that it provides an ideal foundation for the development of end-to-end pipelines, taking into account color evolution and respecting the inhomogeneous nature of astronomical light curve sampling.

astro-ph.IM↗
arXiv2026

Combining astrometry with pulsar timing: the first joint analysis of very low frequency gravitational waves

The pHz to sub-nHz GW regime remains largely unexplored but is crucial for mapping the early inspiral stage of SMBHBs and probing early-Universe physics. Astrometry and pulsar timing offer orthogonal and deeply complementary secular observables to investigate this frequency band. We aim to present the first joint data analysis combining real astrometric proper motions with binary pulsar timing, to search for and constrain continuous gravitational waves (CWs) sourced by SMBHBs in the ultra-low-frequency regime ($10^{-12} \le f_{GW} \le10^{-9} Hz$). We employ a Bayesian model selection and upper limit estimation framework to combine apparent proper motion displacements of $\sim 1.5 \times 10^6$ quasars from the Gaia CRF3 catalog with the line-of-sight orbital period derivatives ($\dot{P}_b$) of 11 high-precision binary pulsars. To prevent spurious detections, we heavily model instrumental and astrophysical systematics: we propagate the Galactic potential uncertainty for pulsars via Monte Carlo simulations and perform a Vector Spherical Harmonics (VSH) decomposition up to the octupole order (l=3) for quasars. We find no statistically significant evidence for a CWs signal in the joint analysis ($\ln B_{joint} = -0.42 \pm 0.03$). In the absence of detection, we set the tightest constraints to date on CW strain in the pHz band, yielding a 95% upper limit of $h_0\le6.4x10^{-11}$ at a reference frequency of $f_{ref} = 4 \times 10^{-10}$ Hz. The combined dataset achieves full sky coverage and improves single-dataset upper limits by 20%-30%. Combining orthogonal observables successfully breaks spatial degeneracies intrinsic to isolated searches. Furthermore, forecasts from inj.-rec. indicate that with the extended temporal baseline and reduced uncertainties of the upcoming Gaia DR4, this joint framework is poised to break the $h_0 < 10^{-11}$ upper limit barrier for sub-nHz CWs.

astro-ph.IM↗