Home
Research
Publications

Resume
Courses
Links

Astronomy
Travel

Semi-Supervised Learning
    Can unlabeled data be used to train a classifier?  Traditional classifiers need labeled data (feature / label pairs) to train.  Labeled instances however are often difficult, expensive, or time consuming to obtain, as they require the efforts of experienced human annotators.  Meanwhile unlabeled data may be relatively easy to collect, but there has been few ways to use them.  Semi-supervised learning addresses this problem by using large amount of unlabeled data, together with the labeled data, to build better classifiers.  Because semi-supervised learning requires less human effort and gives higher accuracy,  it is of great interest both in theory and in practice.
Statistical Language Modeling
Multimodal User Interfaces