Research map / Bandits and online learning / Contextual bandits and Thompson sampling
CBRAP: Contextual Bandits with RAndom Projection
Xiaotian Yu, Michael R. Lyu, Irwin King
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 changed | Where | Measured | Effect |
|---|---|---|---|
| project contexts with random projection (CBRAP.RS m=50) vs full-dimension LinUCB | linear contextual bandit · s4 (n=10000) | cumulative payoff: 967.7 -> 998.1 | Improved |
| CBRAP.RS vs LinUCB | linear contextual bandit · s1 (n=1000) | cumulative payoff: 989.7 -> 994.6 | Improved |
| CBRAP vs BallExp and SLUCB | high-dimensional bandits · s1-s4 | cumulative payoff: 206.8 -> 994.6 (s1 vs BallExp) | Improved |
| CBRAP.SG instead of LinUCB | linear contextual bandit · MovieLens | cumulative payoff: 2595.2 -> 2693.0 | Improved |
| CBRAP.RS (m=20) instead of LinUCB | linear contextual bandit · MovieLens | cumulative payoff: 2595.2 -> 2303.3 | Hurt |
| random projection (CBRAP.RS m=20) instead of full-dimension LinUCB | linear contextual bandit · s1 | time (s) | Improved |
| increase reduced dimension from 20 to 50 (Gaussian matrix) (ablation) | CBRAP.SG · s1 | cumulative payoff: 981.4 -> 955.6 | Hurt |
| increase reduced dimension from 20 to 50 (random sign matrix) (ablation) | CBRAP.RS · s2 | cumulative payoff: 996.3 -> 998.8 | No clear effect |
| increase reduced dimension from 20 to 50 (Gaussian matrix) (ablation) | CBRAP.SG · MovieLens | cumulative payoff: 2693.0 -> 2302.1 | Hurt |
| increase reduced dimension from 20 to 50 (random sign matrix) (ablation) | CBRAP.RS · Jester | cumulative payoff: 1028.0 -> 1065.1 | Improved |
| increase sparsity of contexts (ablation) | CBRAP · s1 | cumulative payoff | No clear effect |
Closest papers by meaning
- Contextual Bandits with Packing and Covering Constraints: A Modular Lagrangian Approach via Regression (JMLR 2024 · similarity 0.90)
- High-dimensional Contextual Bandit Problem without Sparsity (NeurIPS 2023 · similarity 0.90)
- High-dimensional Linear Bandits with Knapsacks (ICML 2024 · similarity 0.90)
- Fast and Sample Efficient Multi-Task Representation Learning in Stochastic Contextual Bandits (ICML 2024 · similarity 0.89)
- Conservative Contextual Bandits: Beyond Linear Representations (ICLR 2025 · similarity 0.89)
- Contextual Bandits for Unbounded Context Distributions (ICML 2025 · similarity 0.88)
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.
