Discriminatively regularized least-squares classification
Discriminatively regularized least-squares classification
复制标题
判别正则化最小二乘分类
DOI：
10.1016/j.patcog.2008.07.010
复制
发表时间：
2009
影响因子：
8
通讯作者：
Yang, Qiang
中科院分区：
文献类型：
--
作者：
Xue, Hui;Chen, Songcan;Yang, Qiang
在过去的几十年里，正则化理论被广泛应用于机器学习的各个领域，衍生出了一大类新的算法。传统上，正则化只注重平滑，没有充分利用对分类至关重要的底层区分知识。本文提出了一种新的最小二乘正则化算法，称为区分正则化最小二乘分类(DRLSC)方法，它是专门为分类而设计的。在几种新的几何激励方法的启发下，DRLSC直接将判别信息和样本的局部几何信息嵌入到正则化项中，以尽可能多地挖掘样本内部的潜在知识，目标是最大化每个局部区域内不同类别样本之间的差值。此外，通过在公式中嵌入等式类型的约束，DRLSC的解可以从求解一组线性方程组而来，并且框架自然包含多类问题。在玩具和现实问题上的实验表明，DRLSC在分类性能上往往优于经典的正则化算法，包括正则化网络、支持向量机和最近研究的一些流形正则化技术。
Over the past decades, regularization theory is widely applied in various areas of machine learning to derive a large family of novel algorithms. Traditionally, regularization focuses on smoothing only, and does not fully utilize the underlying discriminative knowledge which is vital for classification. In this paper, we propose a novel regularization algorithm in the least-squares sense, called discriminatively regularized least-squares classification (DRLSC) method, which is specifically designed for classification. Inspired by several new geometrically motivated methods, DRLSC directly embeds the discriminative information as well as the local geometry of the samples into the regularization term so that it can explore as much underlying knowledge inside the samples as possible and aim to maximize the margins between the samples of different classes in each local area. Furthermore, by embedding equality type constraints in the formulation, the solutions of DRLSC can follow from solving a set of linear equations and the framework naturally contains multi-class problems. Experiments on both toy and real world problems demonstrate that DRLSC is often superior in classification performance to the classical regularization algorithms, including regularization networks, support vector machines and some of the recent studied manifold regularization techniques.
登录
查看更多内容
DOI：
--
发表时间：
2004
期刊：
--
影响因子：
--
作者：
M. Belkin;P. Niyogi;Vikas Sindhwani
通讯作者：
M. Belkin;P. Niyogi;Vikas Sindhwani
DOI：
10.1109/tc.1978.1674981
发表时间：
2015
期刊：
--
影响因子：
--
作者：
通讯作者：
--
DOI：
10.1145/1273496.1273502
发表时间：
2007-06
期刊：
--
影响因子：
--
作者：
S. Asharaf;M. Murty;S. Shevade
通讯作者：
S. Asharaf;M. Murty;S. Shevade
DOI：
--
发表时间：
2005
期刊：
--
影响因子：
--
作者：
S. Szedmák;J. Shawe-Taylor
通讯作者：
S. Szedmák;J. Shawe-Taylor
DOI：
--
发表时间：
2005
期刊：
--
影响因子：
--
作者：
通讯作者：
--