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:352@isdm.umontpellier.fr
DTSTART;TZID=Europe/Paris:20250527T140000
DTEND;TZID=Europe/Paris:20250527T140000
DTSTAMP:20260828T093627Z
URL:https://isdm.umontpellier.fr/events/vertical-federated-learning-with-m
 issing-features-during-training-and-inference-3/
SUMMARY:Vertical Federated Learning with Missing Features During Training a
 nd Inference
DESCRIPTION:Online\n\nMachine Learning in Montpellier\, Theory &amp\; Pract
 ice - Pedro Valdeira (Carnegie Mellon)\n\nVertical federated learning trai
 ns models from feature-partitioned datasets across multiple clients\, who 
 collaborate without sharing their local data. Standard approaches assume t
 hat all feature partitions are available during both training and inferenc
 e. Yet\, in practice\, this assumption rarely holds\, as for many samples 
 only a subset of the clients observe their partition. However\, not utiliz
 ing incomplete samples during training harms generalization\, and not supp
 orting them during inference limits the utility of the model. Moreover\, i
 f any client leaves the federation after training\, its partition becomes 
 unavailable\, rendering the learned model unusable. Missing feature blocks
  are therefore a key challenge limiting the applicability of vertical fede
 rated learning in real-world scenarios. To address this\, we propose LASER
  VFL\, a vertical federated learning method for efficient training and inf
 erence of split neural network-based models that is capable of handling ar
 bitrary sets of partitions. Our approach is simple yet effective\, relying
  on the sharing of model parameters and on task-sampling to train a family
  of predictors. We show that LASER-VFL achieves a convergence rate for non
 convex objectives and\, under the Polyak-Łojasiewicz inequality\, it achi
 eves linear convergence to a neighborhood of the optimum. Numerical experi
 ments show improved performance of LASER-VFL over the baselines. Remarkabl
 y\, this is the case even in the absence of missing features. For example\
 , for CIFAR-100\, we see an improvement in accuracy of % when each of four
  feature blocks is observed with a probability of 0.5 and of % when all fe
 atures are observed. The code for this work is available at https://isdm.u
 montpellier.fr/github.com/Valdeira/LASER-VFL.\n\nMachine Learning in Montp
 ellier\, 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