Loss is its own Reward: Self-Supervision for Reinforcement Learning.
Evan ShelhamerParsa MahmoudiehMax ArgusTrevor DarrellPublished in: ICLR (Workshop) (2017)
Keyphrases
- reinforcement learning
- state space
- function approximation
- reward function
- eligibility traces
- reinforcement learning algorithms
- active learning
- learning algorithm
- model free
- markov decision processes
- partially observable environments
- policy gradient
- optimal control
- learning process
- multi agent
- machine learning
- optimal policy
- dynamic programming
- reinforcement learning methods
- data sets
- learning tasks
- transfer learning
- temporal difference
- learning agent
- control policy
- average reward
- markov decision problems
- neural network