Conveners
Machine learning in astronomy: AGN, transient events, cosmology and others: Friday block 1
- Yu Wang (ICRA/ICRANet/INAF)
- Fatemeh Rastegar Nia (Alzahra university and ICRANet)
- Narek Sahakyan (ICRANet-Armenia)
- Rahim Moradi (ICRANet and ICRA-Sapienza)
Machine learning in astronomy: AGN, transient events, cosmology and others: Friday block 2
- Narek Sahakyan (ICRANet-Armenia)
- Yu Wang (ICRA/ICRANet/INAF)
- Fatemeh Rastegar Nia (Alzahra university and ICRANet)
- Rahim Moradi (ICRANet and ICRA-Sapienza)
Description
The increasing adoption of machine learning (ML) and deep learning (DL) techniques in astrophysics coincides with the exponential growth in astronomical data volumes driven by advancements in observational technologies and the increasing number of telescopes observing in different bands. This surge in data has not only revolutionized our approach to studying the universe but has also led to the development of sophisticated ML/DL tools capable of handling large datasets and extracting valuable insights. Through optimization techniques, ML/DL approaches aim to uncover hidden characteristics within data, enabling faster simulations, improved observations, and deeper understandings of cosmic phenomena. This parallel session aims to provide a comprehensive overview and showcase the latest developments in ML/DL applications in astrophysics. Topics covered include galaxy/star/quasar classification, redshift analysis, estimation of stellar atmospheric properties, simulation of vast cosmic structures, identification of reionization origins, detection of transient events, differentiation of gravitational lensing impacts, reconstruction of cosmic microwave background signals, modeling the observed data, exoplanets discovery and model derivation.
Gamma-ray Bursts (GRBs) are one of the most energetic phenomena in the cosmos, whose study probes physics beyond the reach of laboratories on Earth. Yet, our quest to fully unravel the origin of these events and comprehend their underlying physics is far from complete. Central to this pursuit is the rapid classification of GRBs to guide follow-up observations and analysis across the...
Constraining cosmological parameters for galaxy clustering analyses using the three-point correlation function, despite being pivotal, has historically been limited by the high computational cost of modelling. Here, we introduce a new emulator, based on a convolutional neural network, developed within the framework of a Euclid Preparation Key-Project activity, which substantially accelerates...
Gravitational-wave detections and open public alerts enabled prompt multimessenger studies for the global community. There is an ongoing effort to assimilate and invent machine learning techniques that will allow faster and more confident detections. Better characterized detections of more gravitational-wave events thereby will expand multimessenger science. I will highlight trailblazing...
Identifying gamma-rays and rejecting the background of cosmic ray hadrons are crucial for very-high-energy gamma-ray observation and relevant scientific research. Based on the simulated data from the square kilometer array (KM2A) of LHAASO, eight high-level features are extracted for the gamma/hadron classification. Machine-learning (ML) models, including logistic regression, support vector...
Quasar absorption line is a powerful tool for studying the universe, enabling us to probe distant gas, dust, and galaxy formation and evolution. However, detecting Ca II absorbers is particularly challenging, requiring significant time and effort. Existing deep learning methods often produce a high number of false positives and still require extensive manual verification, achieving an...
I will discuss the use of machine learning and more specifically of convolutional neural network (CNN) to enable fitting of blazar (or other objects) SEDs with numerically costly models. In particular, I will describe the necessary ingredients, the numerical approach and tools, the setup of the neural network and the training steps. Finally, I will discuss future plans and improvements.
The search for gravitational wave signals in the data collected by the current ground-based interferometers is a complex problem, especially when only one detector operates. Modern deep learning approaches could contribute to find a solution. I'll discuss the detection problem and present the work detailed in https://iopscience.iop.org/article/10.1088/1361-6382/ad40f0 where we investigate the...
Unmodeled gravitational-wave signals from magnetars are expected to be weak and challenging to detect in LIGO-Virgo-KAGRA data. We introduce a new method to denoise and stack signals from repetitive magnetar bursts, such as the 2020 SGR 1935+2154 burst storm which produced 217 bursts in 1120 seconds. Our method involves identifying bursts in electromagnetic data and searching for corresponding...
