DeepSynth: Automata Synthesis for Automatic Task Segmentation in Deep Reinforcement Learning.
Mohammadhosein HasanbeigNatasha Yogananda JeppuAlessandro AbateTom MelhamDaniel KroeningPublished in: AAAI (2021)
Keyphrases
- reinforcement learning
- fully automatic
- image segmentation
- level set
- segmentation algorithm
- statistical classification
- semi automatic
- medical images
- multiscale
- grey level
- object segmentation
- region growing
- automatic image analysis
- model free
- pixel level
- function approximation
- cellular automata
- markov decision processes
- learning algorithm
- program synthesis
- segmentation errors
- features extraction
- finite automata
- state space
- finite state
- shape prior
- machine learning
- segmentation method
- dynamic programming
- multi agent
- segmentation accuracy
- active contours
- gray level
- tree automata
- optic disc
- image processing
- probabilistic automata
- symbol recognition
- lesion segmentation