Optimal Regret Is Achievable with Bounded Approximate Inference Error: An Enhanced Bayesian Upper Confidence Bound Framework.
Ziyi HuangHenry LamAmirhossein MeisamiHaofeng ZhangPublished in: NeurIPS (2023)
Keyphrases
- approximate inference
- upper confidence bound
- graphical models
- bayesian networks
- probabilistic model
- contextual bandit
- probabilistic inference
- belief propagation
- message passing
- latent variables
- conditional random fields
- worst case
- dynamic programming
- support vector
- parameter estimation
- probability distribution
- markov chain monte carlo
- variational methods
- free energy
- natural language