Robust and Scalable Conversational AI

Yun-Nung Chen

Abstract: Even conversational systems have attracted a lot of attention recently, the current systems sometimes fail due to the errors from different components. This keynote talk presents following research directions for improving robustness and scalability of conversational systems: 1) we first focus on learning language embeddings specifically for spoken scenarios such as more noisy inputs during inference, and 2) secondly we extend the dialogue systems to access not only structured but unstructured knowledge and propose a novel learning framework for natural language understanding and generation on top of duality towards better scalability. The enhanced robustness and scalability shows the great potential of guiding future research directions.