AI Paper Atlas

Research map / Bandits and online learning / Online convex optimization and games

Algorithms for Average Regret Minimization

Sabine Storandt, Stefan Funke

AAAI 2019 · Bandits and online learning › Online convex optimization and games · approach: Best-of-both-worlds bandit algorithms · 13 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
use heuristic sampling-based volume (eps=0.1) instead of exact volume in greedy (ablation)Greedy average regret selection · RC synthetic, q=16average regret: 2.2 ·10−6 -> 0.0006 (d=3, n=10^6)Hurt
use heuristic sampling-based volume (eps=0.1) instead of exact volume in greedy (ablation)Greedy average regret selection · RC synthetic, q=16running time (s): 1690 -> 0.54 (d=3, n=10^6)Improved
use heuristic instead of exact volume computation (ablation)Greedy average regret selection · Weather d=2average regret: 0.000 -> 0.001 (q=16)No clear effect
decrease sampling eps from 1/20 to 1/40 (ablation)Heuristic greedy · AirData d=7, q=16average regret: 0.1043 -> 0.0715Improved
decrease sampling eps from 1/20 to 1/40 (ablation)Heuristic greedy · AirData d=7, q=16running time (s): 361 -> 18065Hurt
use forward greedy instead of reverse greedy (ablation)Average regret permutation · RC d=4 n=1000accumulated regret R: 0.2790 -> 0.0342Improved
increase k in average k-regret from 1 to 9 (ablation)Heuristic greedy · AirData d=7, q=8average k-regret: 0.0064 -> 0.0001Improved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy