Cornelia Caragea, Doina Caragea, Vasant Honavar
In this paper we address the problem of learning Support Vector Machine (SVM) classifiers from distributed data sources. We identify sufficient statistics for learning SVMs and present an algorithm that learns SVMs from distributed data by iteratively computing the set of sufficient statistics. We prove that our algorithm is exact with respect to its centralized counterpart and efficient in terms of time complexity.
Subjects: 12. Machine Learning and Discovery; 10. Knowledge Acquisition
Submitted: Apr 5, 2005