Multi-domain spoken language understanding with transfer learning

This paper addresses the problem of multi-domain spoken language understanding (SLU) where domain detection and domain-dependent semantic tagging problems are combined. We present a transfer learning approach to the multi-domain SLU problem in which multiple domain-specific data sources can be incorporated. To implement multi-domain SLU with transfer learning, we introduce a triangular-chain structured model. This model effectively learns multiple domains in parallel, and allows use of domain-independent patterns among domains to create a better model for the target domain. We demonstrate that the proposed method outperforms baseline models on dialog data for multi-domain SLU problems.