Anomalus, a DL framework for anomaly detection in Euclid's spectroscopic and photometric data
Pedro Cunha (DIFA - Università di Bologna)
Tuesday 20/10/2026 @ 14:00, Sala Antonio Sollima (IV piano Battiferro)
Ensuring the reliability of spectroscopic datasets is crucial for scientific discovery, particularly in the context of the Euclid mission, which will deliver an increasing volume of spectroscopic data, starting with DR1. This project focuses on automating the detection of anomalies, which in this context one refers to as artefacts, contamination, and other problems during the extraction stage that affected the spectra. By using advanced techniques in deep learning and machine learning, including Variational Autoencoders (VAEs) and ensemble methods, the project aims to identify unusual spectra with high precision.