arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 0903.4817cs.LGcs.CGcs.CVmath.OCstat.ML
收藏

正则化路径复杂性的指数下界

An Exponential Lower Bound on the Complexity of Regularization Paths

Institute of Theoretical Computer Science, ETH Zurich(苏黎世联邦理工学院理论计算机科学研究所);CMAP, École Polytechnique(巴黎综合理工学院应用数学研究中心);INRIA Sophia Antipolis-Méditerranée(法国国家信息与自动化研究所索菲亚·安蒂波利斯-地中海研究中心)

Bernd Gärtner, Martin Jaggi, Clément Maria

摘要:

对于机器学习中的各种正则化优化问题,最近已经开发了计算整个解路径的算法。这些方法中的大多数是由单个参数参数化的二次规划,例如支持向量机(SVM)。解路径算法不仅计算正则化参数的一个特定值的解,而且计算整个解路径,使得最佳参数的选择更容易。已经假设这些分段线性解路径仅具有线性复杂性,即线性地许多弯曲。我们证明,对于支持向量机,在最坏的情况下,这种复杂性可以是指数的训练点的数量。更强地,我们为SVM构建d维中的n个输入点的单个实例,使得至少Θ(2^{n/2})= Θ(2 ^d)随着正则化参数的变化出现许多不同的支持向量子集。

英文摘要:

For a variety of regularized optimization problems in machine learning, algorithms computing the entire solution path have been developed recently. Most of these methods are quadratic programs that are parameterized by a single parameter, as for example the Support Vector Machine (SVM). Solution path algorithms do not only compute the solution for one particular value of the regularization parameter but the entire path of solutions, making the selection of an optimal parameter much easier. It has been assumed that these piecewise linear solution paths have only linear complexity, i.e. linearly many bends. We prove that for the support vector machine this complexity can be exponential in the number of training points in the worst case. More strongly, we construct a single instance of n input points in d dimensions for an SVM such that at least Θ(2^{n/2}) = Θ(2^d) many distinct subsets of support vectors occur as the regularization parameter changes.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑