Conveners
Cosmic Insights from Big Data: How Machine Learning is Decoding the Universe: Thursday block 1
- Giuseppe Angora (INAF Napoli)
- Lorenzo Bazzanini (Universita' degli Studi di Ferrara)
Cosmic Insights from Big Data: How Machine Learning is Decoding the Universe: Thursday block 2
- Giuseppe Angora (INAF Napoli)
- Lorenzo Bazzanini (Universita' degli Studi di Ferrara)
Description
Machine learning (ML) and deep learning (DL) applications in astrophysics have gained enormous momentum in the last few years. Indeed, the exponential growth of astronomical data, thanks to the advancements of observational technologies, has demanded the development of intelligent tools for efficient data handling and extraction of new insights from these vast datasets. ML/DL techniques aim to identify and learn from patterns in data, thereby enhancing simulations and aiding in the understanding of complex phenomena, paving the way for novel discoveries. These techniques have found extensive applications in various domains, including galaxy classification, characterization of galaxy and stellar properties, simulation of large-scale cosmic structures, testing cosmological paradigms, detection of transient events, identification of gravitational lensing effects, and cosmic microwave background inpainting. As the field continues to evolve, interdisciplinary collaboration between astronomers and ML/DL experts will play a crucial role in harnessing the full potential of these techniques to advance our understanding of the universe. This session will delve into these applications and explore the prospects of ML/DL in astrophysics.
Most domains of science are experiencing a paradigm shift due to the advent of a new generation of instruments and detectors which produce data and data streams at an unprecedented rate.
The scientific exploitation of these data, namely Data Driven Discovery, requires interoperability, a massive and optimal use of Artificial Intelligence methods in all steps of the data acquisition,...
This presentation explores some applications of transfer learning in astronomical image analysis, focusing on the usage of a pretrained network (EfficientNet) as a feature extractor. We discuss methods for identifying active galactic nuclei, extracting physical parameters, and detecting anomalies in time series data. Additionally, we present some potential future applications, demonstrating...
Deep learning algorithms have excelled in various domains. Despite this success, few deep-learning models have seen full end-to-end deployment in gravitational-wave searches, both in real-time and on archival data. In particular, there is a lack of standardized software tools for quick implementation and development of novel AI ideas. We address this gap by developing the ML4GW and HERMES...
Gravitational lensing is the relativistic effect generated by massive bodies, which bend the space-time surrounding them. It is a deeply investigated topic in astrophysics and allows validating theoretical relativistic results and studying faint astrophysical objects that would not be visible otherwise. In recent years Machine Learning methods have been applied to support the analysis of the...
Causal discovery techniques have been introduced in the context of machine learning with the goal of finding and constraining causal relations between variables. Multiple algorithms for this task are available, amounting to different operational definitions of causal relations based directly on observational data. These tools have found wide application in disciplines that have limited access...
New wide-field astronomical surveys are opening a new window on the Universe, by collecting data for extraordinarily vaste samples of galaxies. Strong gravitational lenses are rare astronomical events, whose actual number will increase of more than 100 times thanks to the unique data from the Euclid wide survey or the Rubin LSST observations. After finding such 100,000 strong lenses, they need...
We present a proof of concept for an alternative method of strong gravitational lens finding using a conditional Generative Adversarial Network (cGAN). We use Early Release Observation (ERO) images of the Perseus Cluster from Euclid, covering 0. 57sq.degrees on the sky, and the network is based on the pix2pix architecture with an adapted U-Net generator. We train our model to predict...