AI Paper Atlas

Research map / Bandits and online learning / Contextual bandits and Thompson sampling

Deep Bandits Show-Off: Simple and Efficient Exploration with Deep Networks

Rong Zhu, Mattia Rigotti

NeurIPS 2021 · Bandits and online learning › Contextual bandits and Thompson sampling · approach: Thompson sampling analysis · 1 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
add SAU-Sampling exploration to a neural network instead of NeuralLinear posterior samplingNeural-SAU-S · Statlog banditrelative cumulative regret: 1.26 -> 0.62Improved
replace NeuralLinear with Neural-SAU-SNeuralLinear · Mushroom banditrelative cumulative regret: 2.66 -> 2.20Improved
replace NeuralLinear with Neural-SAU-SNeuralLinear · Financial banditrelative cumulative regret: 10.16 -> 5.60Improved
replace NeuralLinear with Neural-SAU-SNeuralLinear · Jester banditrelative cumulative regret: 69.48 -> 61.02Improved
replace NeuralLinear with Neural-SAU-SNeuralLinear · Adult banditrelative cumulative regret: 74.62 -> 78.02Hurt
use SAU-UCB with linear model instead of LinearPosteriorLinearPosterior · Financial banditrelative cumulative regret: 10.77 -> 6.51Improved
use SAU-UCB with linear model instead of LinearPosteriorLinearPosterior · Census banditrelative cumulative regret: 46.70 -> 39.98Improved
use SAU-Sampling versus SAU-UCB with neural network (ablation)Neural-SAU · 7 real-world banditsrelative cumulative regret: 2.20 -> 2.32 (Mushroom)No clear effect
NeuralLinear compared to Neural-SAU variantsNeuralLinear · 7 real-world banditscumulative regret: 32% higher than Neural-SAU-SHurt
use SAU instead of NeuralLinear for runtimeNeural-SAU · 7 real-world banditsexecution time: more than 10 time fasterImproved
use SAU-Neural-S at low delta in Wheel BanditNeuralLinear · Wheel Bandit delta=0.50relative cumulative regret: 18.71 -> 12.49Improved
use SAU-Neural-S at very high delta 0.99NeuralLinear · Wheel Bandit delta=0.99relative cumulative regret: 86.03 -> 106.89Hurt

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