Safe Reinforcement Learning through Meta-learned Instincts.
Djordje GrbicSebastian RisiPublished in: ALIFE (2020)
Keyphrases
- reinforcement learning
- function approximation
- reinforcement learning algorithms
- learning algorithm
- model free
- optimal control
- learned knowledge
- state space
- markov decision processes
- temporal difference learning
- learning process
- multi agent
- policy search
- previously learned
- multi agent reinforcement learning
- function approximators
- machine learning
- unsupervised manner
- temporal difference
- learning problems
- optimal policy
- mobile robot
- dynamic programming
- genetic algorithm