BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//CERN//INDICO//EN
BEGIN:VEVENT
SUMMARY:A Robust Machine Learning Framework for Automated Detection of Sol
 ar Flare Effects in Distributed Magnetometer Networks
DTSTART;VALUE=DATE-TIME:20260820T212000Z
DTEND;VALUE=DATE-TIME:20260820T214000Z
DTSTAMP;VALUE=DATE-TIME:20260828T180853Z
UID:indico-contribution-1419@indico.uni.edu.pe
DESCRIPTION:Speakers: Ricardo Angelo Quispe Mendizábal (Universidad Nacio
 nal Mayor de San Marcos)\nSpace-weather disturbances can disrupt radio com
 munications\, navigation\, defense\, spacecraft operations and modern crit
 ical technological systems\, making reliable characterization of the near-
 Earth response to solar activity important. Distributed magnetometer netwo
 rks continuously monitor this environment\, but heterogeneous data quality
 \, local observing conditions\, and station-by-station analyses limit thei
 r scientific potential. We address this challenge for solar flare effects 
 (SFEs)\, rapid ground magnetic responses to flare-enhanced ionospheric con
 ductivity associated here with X-class flares. We present a scalable and i
 nterpretable machine-learning framework that integrates measurements from 
 six geomagnetic stations while preserving their physical context. The work
 flow combines missing-data characterization\, short-gap imputation\, robus
 t anomaly detection\, baseline removal\, station-wise normalization\, and 
 temporal\, statistical\, spectral\, and multiscale feature extraction. The
 se representations are augmented with flare properties\, cross-station rel
 ationships\, and local solar-geometry descriptors. Four ensemble classifie
 rs are optimized\, probability-calibrated\, and evaluated using class-bala
 nced metrics\, calibration diagnostics\, station-level comparisons\, and c
 ontrolled noise perturbations. The leading calibrated LightGBM model achie
 ves 0.957 balanced accuracy\, an F1 score of 0.943\, a recall of 0.957\, a
 nd a ROC-AUC of 0.978. Interpretable analysis shows that local solar illum
 ination governs SFE detectability\, while magnetic-response features chara
 cterize its strength and evolution. The framework establishes distributed 
 magnetometers as a unified intelligent observatory for scalable\, confiden
 ce-aware space-weather research.\n\nhttps://indico.uni.edu.pe/event/184/co
 ntributions/1419/
LOCATION:UNI\, Lima Perú Auditorio de la Facultad de Ciencias
URL:https://indico.uni.edu.pe/event/184/contributions/1419/
END:VEVENT
END:VCALENDAR