BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Date iCal//NONSGML kigkonsult.se iCalcreator 2.20.2//
METHOD:PUBLISH
X-WR-CALNAME;VALUE=TEXT:Eventi DIAG
BEGIN:VTIMEZONE
TZID:Europe/Paris
BEGIN:STANDARD
DTSTART:20241027T030000
TZOFFSETFROM:+0200
TZOFFSETTO:+0100
RDATE:20251026T030000
TZNAME:CET
END:STANDARD
BEGIN:DAYLIGHT
DTSTART:20250330T020000
TZOFFSETFROM:+0100
TZOFFSETTO:+0200
TZNAME:CEST
END:DAYLIGHT
END:VTIMEZONE
BEGIN:VEVENT
UID:calendar.29359.field_data.0@www.corsodrupal.uniroma1.it
DTSTAMP:20260403T195946Z
CREATED:20250515T131917Z
DESCRIPTION:AbstractMost post-hoc explainability methods for graph classifi
 cation analyze the model’s internal representations rather than explicitly
  capturing its reasoning process. These approaches typically rely on pertu
 rbations\, gradients\, or optimization techniques to infer important featu
 res but do not approximate the decision-making function itself. In this pa
 per\, we propose a novel approach that directly models the GNN’s decision 
 function using a Transparent Explainable Logic Layer (TELL). This logic-ba
 sed approximation enables both instance-level and global-level explanation
 s\, offering insights into how node embeddings contribute to predictions. 
 Unlike conventional methods\, our approach derives explanations that are s
 tructurally aligned with the model’s decision process\, rather than being 
 externally imposed. Through experiments on synthetic and real-world graph 
 classification tasks\, we show that our method produces faithful\, sparse\
 , and stable explanations\, outperforming existing techniques.Speaker Ales
 sio RagnoNSA Lyon - Institut National des Sciences Appliquées de Lyon 
DTSTART;TZID=Europe/Paris:20250522T120000
DTEND;TZID=Europe/Paris:20250522T120000
LAST-MODIFIED:20250515T141506Z
LOCATION:Aula Magna Diag
SUMMARY:Faithful Explanations for Graph Classification using Logic - Alessi
 o Ragno
URL;TYPE=URI:http://www.corsodrupal.uniroma1.it/node/29359
END:VEVENT
END:VCALENDAR