Eluder dimension: localise it!

NeurIPSSpotlight2025

Authors
Alireza Bakhtiari, Alex Ayoub, Samuel McLaughlin Robertson, David Janz, Csaba Szepesvari
Venue
NeurIPS 2025
Track
Spotlight

TL;DR

We establish a lower bound on the eluder dimension in generalised linear model classes, showing that standard eluder dimension-based analysis cannot lead to first-order regret bounds…

Opening excerpt from the authors’ abstract. source

Read the paper

Topics

regret

← All NeurIPS 2025 Spotlight papers · Browse the whole archive