Very Fast Kernel SVM under Budget Constraints
This work addresses efficiency and scalability issues for kernel SVM users in resource-constrained environments, though it appears incremental as it builds on existing clustering and SVM techniques.
The paper tackles the challenge of training kernel SVM models efficiently under computational budget constraints by proposing a fast online algorithm that clusters the input space using LVQ and trains separate kernel SVMs in each cluster with limited support vector sets. The algorithm achieves high accuracy while processing a very high number of samples per second in both training and evaluation.
In this paper we propose a fast online Kernel SVM algorithm under tight budget constraints. We propose to split the input space using LVQ and train a Kernel SVM in each cluster. To allow for online training, we propose to limit the size of the support vector set of each cluster using different strategies. We show in the experiment that our algorithm is able to achieve high accuracy while having a very high number of samples processed per second both in training and in the evaluation.