Research map / Bandits and online learning / Contextual bandits and Thompson sampling
Contextual Gaussian Process Bandits with Neural Networks
Haoting Zhang, Jinghai He, Rhonda Righter, Zuo-Jun Shen, Zeyu Zheng
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 |
|---|---|---|---|
| Replace CGP-UCB additive/multiplicative kernels with NN-AGP-UCB (FCN) | NN-AGP-UCB · Synthetic rewards R1 and R2 | average regret R_T/T | Improved |
| Compare against NeuralUCB and NN-UCB adapted to continuous decisions | NN-AGP-UCB · Synthetic rewards R1 and R2 | average regret R_T/T | Improved |
| Increase NN output dimension m from 2 to 3 to 5 (ablation) | NN-AGP-UCB · Synthetic rewards R1 and R2 | average regret R_T/T | Mixed |
| Apply NN-AGP to rewards that violate the linear assumption g(theta)^T p(x) | NN-AGP-UCB · Synthetic rewards R1 and R2 | average regret R_T/T | Improved |
| Increase input dimension (X dim 2, Theta dim 3 to X dim 5, Theta dim 15) (ablation) | NN-AGP-UCB · Synthetic rewards | regret uncertainty | Hurt |
| Use LSTM for sequence contextual variables instead of CGP-UCB composite kernels | NN-AGP-UCB with LSTM · Queuing problem (theta_t dimension 3 and 10) | cumulative reward | Improved |
| Add a time-series kernel to CGP-UCB (ablation) | CGP-UCB · Queuing problem | cumulative reward | Improved |
| Use GCN for graph contexts instead of CGP-UCB with adjacency or graph kernels | NN-AGP-UCB with GCN · Pricing with diffusion network (5 and 10 nodes) | cumulative reward | Improved |
| Increase network size from 5 to 10 nodes (ablation) | NN-AGP-UCB with GCN · Pricing with diffusion network | advantage over CGP-UCB | Improved |
Closest papers by meaning
- Gaussian Process Bandits for Top-k Recommendations (NeurIPS 2024 · similarity 0.87)
- Improved Regret Analysis in Gaussian Process Bandits: Optimality for Noiseless Reward, RKHS norm, and Non-Stationary Variance (ICML 2025 · similarity 0.87)
- Gaussian Process Upper Confidence Bound Achieves Nearly-Optimal Regret in Noise-Free Gaussian Process Bandits (NeurIPS 2025 · similarity 0.86)
- Contextual Multinomial Logit Bandits with General Value Functions (NeurIPS 2024 · similarity 0.85)
- Contextual Bandits with Online Neural Regression (ICLR 2024 · similarity 0.85)
- Improved Regret Bounds for Gaussian Process Upper Confidence Bound in Bayesian Optimization (NeurIPS 2025 · similarity 0.85)
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.
