一种多重支持向量机在线学习算法的研究

增量式支持向量机学习算法是一种重要的在线学习方法。传统的单增量支持向量机学习算法使用一个数据样本更新支持向量机模型。在增加或删除的数据样本点较多时,这种模型更新模式耗时巨大,具体原因是每个被插入或删除的样本都要进行一次模型参数更新的判断。该文提出一种基于参数规划的多重增量式的支持向量机优化训练算法,使用该训练算法,多重的支持向量机的训练时间大为减少。在合成数据集及真实测试数据集上的实验结果显示,该文提出的方法可以大大降低多重支持向量机训练算法的计算复杂度并提高分类器的精度。

A Multiple Online Support Vector Machine Learning Algorithm

Incremental Support Vector Machines （SVM） is an important online learning method. The conventional training methods utilizes a single training data to update parameters of the SVM model. When there are many training samples that need-ed to be added or removed from the training set, this kind of learning manner is time consuming because each training sample is trained against the model. In this paper, we propose a parameter programming based learning approach to reduce the training time when multiple samples are added into or removed from the learning model. The performance of the proposed algorithm is evaluated in two scenarios：learning with limited resources and active learning. Experimental results indicate that the proposed ap-proach can reduce online SVM training time greatly.