Tuning the Hyperparameters of Anytime Planning: A Metareasoning Approach with Deep Reinforcement Learning.
Abhinav BhatiaJustin SvegliatoSamer B. NashedShlomo ZilbersteinPublished in: ICAPS (2022)
Keyphrases
- hyperparameters
- reinforcement learning
- model selection
- cross validation
- decision theoretic
- bayesian inference
- closed form
- bayesian framework
- random sampling
- support vector
- parameter settings
- prior information
- em algorithm
- noise level
- gaussian process
- maximum likelihood
- sample size
- incomplete data
- anytime algorithms
- maximum a posteriori
- gaussian processes
- incremental learning
- regularization parameter
- dynamic programming
- state space
- markov decision problems
- parameter space
- missing values
- data sets
- expectation maximization
- active learning
- learning process
- lower bound
- image processing
- learning algorithm