Breaking the curse of kernelization: budgeted stochastic gradient descent for large-scale SVM training

By Zhuang Wang, Koby Crammer, Slobodan Vucetic. 2012

200 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Management Science and Operations Research

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Management Science and Operations Research

Citations in the corpus

Sharp analysis of low-rank kernel matrix approximations (2012, neutral): “…algorithms operate online and do not need to compute the full kernel matrix, such as the “forgetron” (Dekel et al., 2005), the “projectron” (Orabona et al., 2008), BGSD (Wang et al., 2012), or LASVM (Bordes et al., 2005), with typically a ﬁxed computational budget and good practical performance.”
Large-scale Online Kernel Learning with Random Feature Reparameterization (2017, methodology): “Their implementations are published as a part of LIBSVM, BudgetedSVM [Wang et al., 2012] and LSOKL [Lu et al.”
A Survey on Ensemble Learning for Data Stream Classification (2017, methodology): “…[John and Langley 1995], like Naive Bayes; decision trees, like Hoeffding Trees [Domingos and Hul-ten 2000]; Stochastic Gradient Descent variations [Wang et al. 2012]; Instance-based (Lazy) methods [Beringer and H¨ullermeier 2007; Zhang et al. 2011a]; and ensemble classiﬁers [Oza 2005; Bifet et…”
Online Learning: A Comprehensive Survey (2018, methodology): “Wang et al. (2012d) proposed a SV merging method called “BSGD+Merge” which replaces the sum of two SV’s αmκ(xm, ·)+αnκ(xn, ·) by a newly created SV αzκ(z, ·), where αm, αn and αz are the corresponding coefficients of xm, xn and z.”
Fast Supervised Hashing with Decision Trees for High-Dimensional Data (2014, methodology): “Even using the stochastic kernel SVM (BSGD) [25] with a support vector budget, the training and testing cost are still very expensive.”