Engagement Rewarded Actor-Critic with Conservative Q-Learning for Speech-Driven Laughter Backchannel Generation.
Öykü Zeynep BayramogluEngin ErzinTevfik Metin SezginYücel YemezPublished in: ICMI (2021)
Keyphrases
- actor critic
- reinforcement learning
- reinforcement learning algorithms
- audio visual
- function approximation
- policy iteration
- optimal control
- temporal difference
- approximate dynamic programming
- neuro fuzzy
- convergence proof
- policy gradient
- gradient method
- model free
- speech recognition
- markov decision processes
- state space
- machine learning
- speech signal
- learning algorithm
- multi agent
- average reward
- finite state
- optimal policy
- monte carlo
- rl algorithms
- linear program
- action selection
- step size
- control system
- partially observable markov decision processes
- cooperative
- recursive least squares
- temporal difference learning
- control strategy
- least squares