Deep Reinforcement Learning-based Dialogue Policy with Graph Convolutional Q-network.
Kai XuZhengyu WangYuxuan LongQiaona ZhaoPublished in: LREC/COLING (2024)
Keyphrases
- reinforcement learning
- optimal policy
- network architecture
- deep learning
- function approximation
- markov decision process
- fully connected
- graphical representation
- action selection
- weighted graph
- machine learning
- spanning tree
- state and action spaces
- policy search
- network structure
- graph structure
- small world
- graph theory
- action space
- complex networks
- state space
- partially observable environments
- mixed initiative
- random graphs
- dialogue system
- edge weights
- link prediction
- markov decision processes
- natural language
- random walk
- peer to peer