Self-attention based deep direct recurrent reinforcement learning with hybrid loss for trading signal generation.
Dongkyu KwakSungyoon ChoiWoojin ChangPublished in: Inf. Sci. (2023)
Keyphrases
- reinforcement learning
- function approximation
- electronic commerce
- signal processing
- reinforcement learning algorithms
- multi agent
- recurrent neural networks
- optimal policy
- machine learning
- autonomous mental development
- frequency domain
- learning algorithm
- state space
- signal detection
- compressive sensing
- non stationary
- model free
- neural network
- learning agents
- deep learning
- heart rate
- action selection
- hybrid learning
- multi agent systems
- robotic systems
- dynamic programming
- visual attention
- supervised learning