Two-stage population based training method for deep reinforcement learning.
Yinda ZhouWeiming LiuBin LiPublished in: HP3C (2019)
Keyphrases
- reinforcement learning
- significant improvement
- high precision
- training process
- experimental evaluation
- high accuracy
- cost function
- dynamic programming
- mobile robot
- synthetic data
- objective function
- feature set
- training phase
- feed forward neural networks
- similarity measure
- classification method
- error rate
- detection method
- clustering method
- semi supervised
- support vector machine
- probabilistic model
- computational cost
- hidden markov models
- prior knowledge