BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//wp-events-plugin.com//7.4.0.1//EN
TZID:Europe/Paris
X-WR-TIMEZONE:Europe/Paris
BEGIN:VEVENT
UID:353@isdm.umontpellier.fr
DTSTART;TZID=Europe/Paris:20250519T140000
DTEND;TZID=Europe/Paris:20250519T140000
DTSTAMP:20260828T093704Z
URL:https://isdm.umontpellier.fr/events/engression-extrapolation-through-t
 he-lens-of-distributional-regression-2/
SUMMARY:Engression: extrapolation through the lens of distributional regres
 sion
DESCRIPTION:Online\n\nMachine Learning in Montpellier\, Theory &amp\; Pract
 ice - Xinwei Shen (ETH Zürich)\n\nDistributional regression aims to estim
 ate the full conditional distribution of a target variable\, given covaria
 tes. Popular methods include linear and tree ensemble based quantile regre
 ssion. We propose a neural networkbased distributional regression methodol
 ogy called ‘engression’. An engression model is generative in the sens
 e that we can sample from the fitted conditional distribution and is also 
 suitable for high-dimensional outcomes. Furthermore\, we find that modelli
 ng the conditional distribution on training data can constrain the fitted 
 function outside of the training support\, which offers a new perspective 
 to the challenging extrapolation problem in nonlinear regression. In parti
 cular\, for ‘preadditive noise’ models\, where noise is added to the c
 ovariates before applying a nonlinear transformation\, we show that engres
 sion can successfully perform extrapolation under some assumptions such as
  monotonicity\, whereas traditional regression approaches such as least-sq
 uares or quantile regression fall short under the same assumptions. Our em
 pirical results\, from both simulated and real data\, validate the effecti
 veness of the engression method. The software implementations of engressio
 n are available in both R and Python.\n\nMachine Learning in Montpellier\,
  Theory &amp\; Practice
ATTACH;FMTTYPE=image/jpeg:https://isdm.umontpellier.fr/wp-content/uploads/
 2026/06/ml-mtp-gC78d5.png
CATEGORIES:ML MTP
END:VEVENT
BEGIN:VTIMEZONE
TZID:Europe/Paris
X-LIC-LOCATION:Europe/Paris
BEGIN:DAYLIGHT
DTSTART:20250330T030000
TZOFFSETFROM:+0100
TZOFFSETTO:+0200
TZNAME:CEST
END:DAYLIGHT
END:VTIMEZONE
END:VCALENDAR