Technical Program

Paper Detail

Presentation #12
Location:Kallirhoe Hall
Session Time:Thursday, December 20, 10:00 - 12:00
Presentation Time:Thursday, December 20, 10:00 - 12:00
Presentation: Poster
Topic: Spoken dialog systems:
Authors: Izzeddin Gur, University of California Santa Barbara, United States; Dilek Hakkani-Tür, Google, United States; Gokhan Tür, Uber AI Labs, United States; Pararth Shah, Facebook, United States
Abstract: In this paper, we introduce end-to-end neural network based models for simulating users in task-oriented dialogue systems. User simulation in dialogue systems is crucial from two different perspectives: (i) automatic evaluation of different dialogue models, and (ii) training individual dialogue system components. We design a hierarchical sequence-tosequence model that first encodes the initial user goal and system turns into fixed length representations using Recurrent Neural Networks (RNN). It then encodes the dialogue history using another RNN layer. At each turn, user responses are decoded from the hidden representations of the dialogue level RNN. This hierarchical user simulator (HUS) approach allows the model to capture undiscovered parts of the user goal without the need of an explicit dialogue state tracking. We further develop several variants by utilizing a latent variable model to inject random variations into user responses to promote diversity and a novel goal regularization mechanism to penalize divergence of user responses from the initial user goal. We evaluate the proposed models on a movie ticket booking domain by systematically interacting each user simulator with various dialogue system policies trained with different objectives and users.