Research

Our work is organized around five themes. Each theme gathers research areas that share a central question.

Map of connections

Follow the threads between themes, areas, projects, people, and publications.

Projects and papers will appear here as they are published.

Research themes

Each theme gathers areas that share a central question.

  1. Perception

    How machines see

    1. Computer Vision

      Visual understanding, image analysis, medical imaging, segmentation, recognition, and generation.

  2. Language

    How machines read, reason, and speak

    1. Language Models & NLP

      Language understanding, generation, retrieval, reasoning, and knowledge-intensive systems.

  3. Junction

    Where modalities meet

    1. Vision-Language Models

      Models that ground language in images and images in language.

    2. Multimodal AI

      Systems that connect visual, textual, and other modalities, with a focus on multimodal understanding and reasoning.

  4. Understanding

    What is learned, and why it holds

    1. Representation Learning

      Deep learning, generative models, evaluation, and new learning methods; the study of what models learn internally.

    2. Causal Inference

      Relationships, interventions, and mechanisms beyond correlation.

  5. Discovery

    Computation in service of science

    1. AI for Science

      Computational intelligence applied to problems in biology, medicine, and other scientific disciplines.

    2. Computational Biology

      Machine learning and computational approaches for biological and biomedical research.