A data-efficient goal-directed deep reinforcement learning method for robot visuomotor skill.
Rong JiangZhipeng WangBin HeYanmin ZhouGang LiZhongpan ZhuPublished in: Neurocomputing (2021)
Keyphrases
- goal directed
- synthetic data
- noisy data
- data sets
- reinforcement learning
- test data
- data analysis
- dynamic programming
- prior knowledge
- input data
- missing data
- statistical methods
- model free
- pairwise
- prior information
- training data
- knowledge discovery
- information loss
- computationally efficient
- autonomous learning
- missing values
- clustering method
- data sources
- preprocessing
- similarity measure
- support vector machine
- image data
- data points
- probabilistic model
- mobile robot
- spectral clustering
- machine learning
- autonomous robots
- data mining
- neural network
- temporal abstractions
- database