BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//CERN//INDICO//EN
BEGIN:VEVENT
SUMMARY:The AI Spectroscopist: Unveiling Galaxy and IGM Physics in the Era
  of Million-Spectrum Surveys
DTSTART;VALUE=DATE-TIME:20260907T094000Z
DTEND;VALUE=DATE-TIME:20260907T101000Z
DTSTAMP;VALUE=DATE-TIME:20260922T191356Z
UID:indico-contribution-2022@events.ncbj.gov.pl
DESCRIPTION:Speakers: Francesco Pistis (National Centre for Nuclear Resear
 ch)\nThe era of wide-field spectroscopic surveys such as WEAVE\, DESI\, an
 d 4MOST\, along with forthcoming facilities like the Wide-Field Spectrosco
 pic Telescope (WST) and MOSAIC\, is delivering millions of spectra that en
 code the chemical and structural history of the Universe. The sheer volume
  of this data makes traditional human-supervised analysis intractable and 
 necessitates the transition to fully automated machine learning (ML) pipel
 ines to efficiently characterize the Intergalactic Medium (IGM)\, the circ
 umgalactic medium (CGM)\, and complex galaxy and AGN physics.\nA fundament
 al challenge in spectroscopic analysis is the accurate estimation of the i
 ntrinsic spectral continuum. We present an optimized autoencoder architect
 ure that achieves superior precision\, with a median Absolute Fractional F
 lux Error (AFFE) of 0.009 for quasar spectra. This model has demonstrated 
 strong generalizability by successfully recovering the Lyα optical depth 
 evolution in unseen DESI data. \nFurthermore\, specialized U-Net architect
 ures are now capable of the generalized detection of metal absorption feat
 ures (e.g.\, CIV\, MgII\, SiIV) with high completeness and purity\, reachi
 ng an F1 score of ≈90% at S/N≈4. Following detection\, physically moti
 vated classifiers allow for robust ion identification with an average 90% 
 accuracy\, reaching ∼100% for reliable doublet systems. This integrated 
 pipeline allows the analysis of ≈10\,000 spectra in just a few seconds.\
 n\nhttps://events.ncbj.gov.pl/event/468/contributions/2022/
LOCATION:
URL:https://events.ncbj.gov.pl/event/468/contributions/2022/
END:VEVENT
END:VCALENDAR