Bipedal walking energy minimization by reinforcement learning with evolving policy parameterization.
Petar KormushevBarkan UgurluSylvain CalinonNikolaos G. TsagarakisDarwin G. CaldwellPublished in: IROS (2011)
Keyphrases
- energy minimization
- reinforcement learning
- optimal policy
- policy search
- graph cuts
- energy function
- action selection
- markov decision process
- markov random field
- global minimum
- markov decision processes
- function approximation
- state space
- image segmentation
- policy gradient
- function approximators
- action space
- reward function
- problems in computer vision
- max flow
- dual decomposition
- reinforcement learning algorithms
- active contour model
- low level vision
- belief propagation
- image dependent
- machine learning
- global minimization
- min cut
- dynamic programming
- weighted constraint satisfaction
- model free
- interactive segmentation
- similarity measure
- early vision
- random fields
- interactive image segmentation
- max flow min cut