Information Fusion

One of my research foci is information fusion. Given the complexity of real-world problems, fusion of information from different sources is very often the only viable way to achieve high performances. I have concentrated on an information-theoretical, trainable fusion method, where the teaching input can be either a regular training set or user feedback. This method provides good results in document processing, and I'm currently trying to show its usefulness in other application domains. My latest experiments and results obtained with this method are listed in my publications. Most of this work was done at the University of Maryland. More recently, I'm experimenting with information fusion for biological applications at the CAS-MPG Partner Institute for Computational Biology in Shanghai.

• Stefan Jaeger, Huanfeng Ma, David Doermann. Combining Classifiers with Informational Confidence.
  In: Machine Learning in Document Analysis and Recognition, Editors: Simone Marinai, Hiromichi Fujisawa, Springer Lecture Notes in Computer Science, 2008.
  

Group of Systems Bioinformatics

Language and Media Processing Laboratory