My research group investigates how humans understand and use data. The core techniques we use include representation learning approaches including neuro-symbolic techniques using probabilistic graphical models, Transformer-based language models, and graph neural networks. Our projects include language modeling techniques that use human cognition to model numerical understanding, commonsense reasoning, and biases, task-focused dialogue agents that focus on user-centric measures of utility, and table understanding systems that dissect spreadsheets and summarize them in human-understandable ways.

Before applying for this opportunity you need to submit your online profile. Click the button below to continue.