My research focuses on natural language processing and artificial intelligence with a particular interest in the field of machine learning for natural language dialogue processing (in both spoken and written forms). In natural language dialogue two or more participants interact in order to exchange information, perform a task, negotiate, or even for pure entertainment purposes. Natural language dialogue systems allow users to interact with computers using language (spoken or written) by trying to simulate how humans interact with one another. Due to the technological challenges they impose as well as their potential for commercial exploitation, natural language dialogue systems have attracted increased attention in both academia and industry. Natural language dialogue systems have applications in many different domains, for example, providing information, command-and-control, tutorial dialogue (and generally dialogue for training, e.g., leadership skills, interviewing skills, etc.), healthcare applications (e.g., virtual human interviewers that interact with people who suffer from depression or post-traumatic stress), storytelling, controlling smart homes, companions for the elderly, etc.

The underlying concept behind my research work is how to learn human behavior patterns from existing human-machine interaction data in order to make natural language dialogue systems more human-like, robust to errors and misunderstandings, and adaptive to different types of human users and domains. This is an iterative procedure: build a dialogue system, have the system interact with humans, analyze the data from these interactions, use the data from these interactions to improve the dialogue system (e.g., by learning behavior patterns directly from the data), and so forth.

My research interests include all aspects of natural language dialogue processing with a focus on statistical dialogue management (reinforcement learning of dialogue policies), expressive conversational speech synthesis, and speech recognition.