Abstract | ||
---|---|---|
The combination of Support Vector Machines with very high dimensional kernels, such as string or tree kernels, suffers from two major drawbacks: first, the implicit representation of feature spaces does not allow us to understand which features actually triggered the generalization; second, the resulting computational burden may in some cases render unfeasible to use large data sets for training. We propose an approach based on feature space reverse engineering to tackle both problems. Our experiments with Tree Kernels on a Semantic Role Labeling data set show that the proposed approach can drastically reduce the computational footprint while yielding almost unaffected accuracy. |
Year | Venue | Keywords |
---|---|---|
2009 | CoNLL | efficient linearization,computational burden,cases render unfeasible,large data set,tree kernels,feature space reverse engineering,feature space,tree kernel function,support vector machines,computational footprint,semantic role,kernel function,support vector machine,reverse engineering |
Field | DocType | Citations |
Feature vector,Data set,Computer science,Reverse engineering,Support vector machine,Tree kernel,Footprint,Artificial intelligence,Machine learning,Linearization,Semantic role labeling | Conference | 7 |
PageRank | References | Authors |
0.49 | 27 | 2 |
Name | Order | Citations | PageRank |
---|---|---|---|
daniele pighin | 1 | 289 | 18.72 |
Alessandro Moschitti | 2 | 3262 | 177.68 |