A Monte Carlo EM Approach for Partially Observable Diffusion Processes: Theory and Applications to Neural Networks.
Javier R. MovellanPaul MineiroRuth J. WilliamsPublished in: Neural Comput. (2002)
Keyphrases
- monte carlo
- partially observable
- diffusion processes
- markov chain
- state space
- reinforcement learning
- diffusion process
- decision problems
- markov decision processes
- dynamical systems
- partial differential equations
- expectation maximization
- temporal difference
- image processing
- markov decision problems
- infinite horizon
- generative model
- reward function
- information diffusion
- decision making
- learning algorithm
- denoising
- probabilistic model
- dynamic programming
- search space
- search algorithm