Output details
15 - General Engineering
Brunel University London
An ontology enhanced parallel SVM for scalable spam filter training
Machine learning techniques such as Support Vector Machines (SVMs) are facing an increasing challenge in dealing with big data. The MapReduce programming model has become the de facto computing model in support of data intensive applications in cloud computing systems. To speed up the computation process in SVM training, this paper parallelises SVM with the MapReduce model. More importantly, it introduces domain knowledge to minimise accuracy degradation of the parallelised SVM in classification. This is a pioneer work in this field which has partially led to a successful EPSRC project (EP/K006487/1) on big data analytics for smart grids.