An Improved Dyna-Q Algorithm Based in Reverse Model Learning.
Yi-Jia TsengKao-Shing HwangWei-Cheng JiangTsung-Chuan HuangSong-Shyong ChenPublished in: ICSSE (2015)
Keyphrases
- learning algorithm
- probabilistic model
- learning phase
- algorithm employs
- learning scheme
- mathematical model
- automatically learned
- cost function
- recognition algorithm
- objective function
- parameter estimation
- input data
- theoretical analysis
- learned models
- learning process
- prior knowledge
- dynamic programming
- tree structure
- similarity measure
- selection algorithm
- optimization algorithm
- np hard
- expectation maximization
- classification algorithm
- k means
- learning mechanism
- estimation algorithm
- detection algorithm
- multiple kernel
- fully connected
- optimization model
- kalman filter
- monte carlo
- learning models
- simulated annealing
- maximum likelihood
- bayesian framework
- active learning
- search space
- temporal difference learning
- pac model
- em algorithm
- optimal solution