Our research spans several high-impact areas of AI, from understanding and improving human language to developing systems that interact with the world in multiple ways. Below are the key areas we are currently exploring:
Our NLP research focuses on advancing AI for low-resource languages by developing models that can perform well with limited data. We also explore in-context learning, enabling models to adapt to new tasks without retraining. Additionally, we investigate causality in language models, helping models better understand and generate language based on cause-and-effect relationships.
Our research in Multimodal AI focuses on integrating diverse data modalities, such as text, images, and video. A key area of interest is addressing the representation and alignment problem in multimodal/multisensory AI, ensuring that information from different modalities is effectively represented and aligned to achieve seamless understanding and decision-making across inputs.