Research Interests
My Intelligent Adaptive Interventions research group aims to transform everyday technology for people into intelligent, perpetually improving systems, by integrating human intelligence with statistical machine learning. Simple examples in education include enhancing the explanations students receive for how to solve problems, and examples in health include personalizing which text messages get people to exercise. Our approach is to deploy randomized A/B comparisons that compare alternative actions, creating these A/B comparisons using insights from human-computer interaction and experimental psychology, as well as harnessing ideas from users and designers through crowdsourcing or 'human computation' to perpetually expand the set of actions/conditions being tested. We then apply algorithms to discover which actions are effective (and for which subgroups of users) and rapidly provide the most effective actions to future users, to enhance and personalize their experience. For example, existing research and active collaborations apply and extend algorithms in reinforcement learning (e.g. multi-armed contextual bandits) along with models from statistics. As shown in the diversity of publication venues, my group bridges research in human-computer interaction, experimental psychology, applied statistics and machine learning.