Georgia Institute of Technology
My research focuses on learning multimodal representations and generating multimodal responses to augment large language models for human-machine dialogue. I am interested in extending current conversational language models to understand information from textual and non-textual sources such as tabular data, images, and videos. Furthermore, I am also interested in machine learning techniques to generate textual and non-textual information. Ultimately, I hope to show that equipping models with the ability to understand multiple modalities enhances their ability to generate multimodal responses, and conversely, building models that generate multimodal conversational responses helps them create better representations.