Name
Abstract | ||
|---|---|---|
Modelling dialogue as a Partially Observable Markov Decision Process (POMDP) enables a dialogue policy robust to speech understanding errors to be learnt. However, a major challenge in POMDP policy learning is to maintain tractability, so the use of approximation is inevitable. We propose applying Gaussian Processes in Reinforcement learning of optimal POMDP dialogue policies, in order (1) to make the learning process faster and (2) to obtain an estimate of the uncertainty of the approximation. We first demonstrate the idea on a simple voice mail dialogue task and then apply this method to a real-world tourist information dialogue task. |
Year | Venue | Keywords |
|---|---|---|
2010 | SIGDIAL Conference | speech understanding error,gaussian processes,pomdp-based dialogue manager,real-world tourist information dialogue,optimal pomdp dialogue policy,modelling dialogue,major challenge,fast policy optimisation,partially observable markov decision,pomdp policy learning,simple voice mail dialogue,dialogue policy,gaussian process |
Field | DocType | Citations |
Partially observable Markov decision process,Policy learning,Computer science,Gaussian process,Artificial intelligence,Machine learning,Reinforcement learning | Conference | 32 |
PageRank | References | Authors |
1.45 | 3 | 7 |
Name | Order | Citations | PageRank |
|---|---|---|---|
| M. Gašić | 1 | 38 | 2.40 |
| F. Jurčíček | 2 | 35 | 1.90 |
| S. Keizer | 3 | 35 | 1.90 |
| F. Mairesse | 4 | 84 | 5.44 |
| B. Thomson | 5 | 55 | 3.63 |
| Kai Yu | 6 | 1082 | 90.58 |
| S. Young | 7 | 134 | 7.84 |