Daphne Koller
Location: (Stanford, CA)
Personal Research Web Page: http://ai.stanford.edu/~koller/
Keywords: statistical machine learning, probabilistic models, computer vision, computational systems biology
Posted on: Sunday, May 31st, 2009
Broad Research Area: AI / Machine Learning / Robotics / Vision, Scientific/Medical Informatics
Research Interests:
I am interested in developing and using machine learning and probabilistic methods to derive high-level representations from large amounts of noisy data. I work both on developing basic methods and on real-world applications, using each to inspire and support the other. In terms of applications, I have two main thrusts. The first is understanding the physical world, in particular extracting symbolic representations of scenes from image data and other modalities (such as laser range). The second is understanding biological systems from different types of data: genomic, imaging, and (more recently) clinical. In terms of foundational developments, I am interested in inference and learning in more complex probabilistic models, in learning representations that are more richly structured, and in learning from weak or indirect supervision (for example, transfer learning). Representative projects over the past few years include, for example, using context in holistic scene understanding, understanding how individual genetic variation can affect gene regulation and phenotype, and using prior knowledge about the relationships between features and tasks to learn informed priors for regression problems.
