Learning Intelligent Assistants for Understanding Behavior

Can we build learning systems that sift through transaction data to understand behavior and improve via the interaction with domain experts? We believe so!

This talk will analyze how, by formulating adequately our research questions and by employing models that are capable of capturing domain interaction, learning can be made interactive and scalable. Next, we will discuss how inverse reinforcement learning (IRL) and can be employed for assisting experts in surveillance tasks such as intent recognition and detecting abnormal behavior. Finally, we will outline open research questions for usable expert-interactive learning.