Helping an Agent Reach a Different Goal by Action Transfer in Reinforcement Learning.
Yuchen WangFenghui RenMinjie ZhangPublished in: Australasian Conference on Artificial Intelligence (2019)
Keyphrases
- agent learns
- reinforcement learning
- action selection
- reward signal
- transfer learning
- state action
- reward shaping
- multi agent
- state space
- action space
- learning agent
- agent receives
- function approximation
- learning algorithm
- human users
- autonomous agents
- knowledge transfer
- reinforcement learning algorithms
- mobile robot
- machine learning
- model free
- multi agent systems
- internal state
- plan library
- temporal difference
- multiagent systems
- optimal policy
- markov decision process
- joint action
- function approximators
- initially unknown
- partially observable domains
- partial observations
- transition model
- action selection mechanism
- autonomous learning
- previously learned
- sensory inputs
- action models
- learning agents
- plan execution
- partially observable
- multiple agents
- pedagogical agents
- software agents
- intelligent agents
- learning process