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
TZNAME:CET
END:STANDARD
BEGIN:DAYLIGHT
DTSTART:20250330T020000
TZOFFSETFROM:+0100
TZOFFSETTO:+0200
TZNAME:CEST
END:DAYLIGHT
END:VTIMEZONE
BEGIN:VEVENT
UID:calendar.28683.field_data.0@oba.diag.uniroma1.it
DTSTAMP:20260407T215536Z
CREATED:20250312T065547Z
DESCRIPTION:TitleDebiasing SHAP scores in tree ensembles Abstract: Our rese
 arch investigates a bias inherent in the SHapley Additive exPla-nations (S
 HAP) values used for explaining the results of tree based machine learning
  models. This bias leads to models being overly sensitive to features with
  highentropy\, resulting in inflated importance scores for these features.
  We propose a novel method called 'shrunk SHAP' to address this issue by s
 eparating the model training and prediction processes and comparing the re
 sulting SHAP values\, thus reducing the bias towards high-entropy features
  and providing more accurate explanations. Our algorithm is also able to e
 nable the detection of overfitting issues at the feature level. The effect
 iveness of the method is demonstrated through simulations and real-world e
 xamples\, highlighting the potential of 'shrunk SHAP' for improving the in
 terpretability of random forest and boosted tree models. Online link: http
 s://sony-research.zoom.us/j/86098451996?pwd=i0ayxO9EhkanwLSahb4uww1IBTWiZT
 .1 Meeting ID: 860 9845 1996 Passcode: 183496 Organized by Prof. Roberto C
 apobianco Reference: https://link.springer.com/article/10.1007/s10182-023-
 00479-7
DTSTART;TZID=Europe/Paris:20250321T093000
DTEND;TZID=Europe/Paris:20250321T093000
LAST-MODIFIED:20250312T070532Z
LOCATION:Aula Magna\, DIAG
SUMMARY:Seminar by Markus Locker: Debiasing SHAP scores in tree ensembles. 
 - Markus Löcher
URL;TYPE=URI:http://oba.diag.uniroma1.it/node/28683
END:VEVENT
END:VCALENDAR