APD: Learning Diverse Behaviors for Reinforcement Learning Through Unsupervised Active Pre-Training.
Kailin ZengQiyuan ZhangBin ChenBin LiangJun YangPublished in: IEEE Robotics Autom. Lett. (2022)
Keyphrases
- reinforcement learning
- supervised learning
- learning process
- learning problems
- learning algorithm
- unsupervised learning
- deep architectures
- learning speed
- learning systems
- active learning
- prior knowledge
- training set
- function approximation
- weakly supervised
- semi supervised
- learning capabilities
- neural network
- online training
- learning stage
- temporal difference learning
- learning agents
- training data
- robot control
- structured prediction
- feedforward neural networks
- training process
- training examples
- online learning
- state space