This paper may be useful to you. They describe how to use labelled and unlabelled data with mixture discriminant analysis and the classification EM algorithm.

http://onlinelibrary.wiley.com/doi/10.1111/j.1467-9876.2005.00526.x/abstract;jsessionid=C47886E0D3FBFB69C2F7674165C31882.d01t03?deniedAccessCustomisedMessage=&userIsAuthenticated=false

Alternatively you may want to look at semi-supervised classification methods from the machine learning literature.