Statistical learning theory

About: Statistical learning theory is a research topic. Over the lifetime, 1618 publications have been published within this topic receiving 158033 citations.

Support Vector Machine Multi-class Classification Based on an Improved Cross Validation Algorithm

Abstract: How to find the optimal parameters of SVM to train the model in order to have a better recognition rate when treating unknown test samples has been the most important issue to solve practical problems.Traditional cross validation algorithm only find the optimal parameters in the whole scale of values.A new algorithm was proposed which take into account the extreme values in the limited scale.Results show that this algorithm has a better recognition rate.

Accelerating SVM on large scale regression problem

Abstract: The support vector machine (SVM) is a new generation learning system based on recent advances in statistical learning theory. There is evidence showing that SVMs can deliver state-of-the art performance in real-world applications such as text categorisation, handwritten character recognition, image classification, biosequence analysis, etc. In this paper, we investigated the large scale regression problem by using SVMs, but instead of solving quadratic programming slowly, we try to accelerate the regression speed by linear programming formulation using 1-norm or /spl infin/-norm for the model complexity. Computational experiments show that linear programming gives us very good performance without deteriorating the results.

A Note on Perturbation Results for Learning Empirical Operators

Abstract: A large number of learning algorithms, for example, spectral clustering, kernel Principal Components Analysis and many manifold methods are based on estimating eigenvalues and eigenfunctions of operators defined by a similarity function or a kernel, given empirical data. Thus for the analysis of algorithms, it is an important problem to be able to assess the quality of such approximations. The contribution of our paper is two-fold: 1. We use a technique based on a concentration inequality for Hilbert spaces to provide new much simplified proofs for a number of results in spectral approximation. 2. Using these methods we provide several new results for estimating spectral properties of the graph Laplacian operator extending and strengthening results from [26].

Study on Support Vector Machine in Calculating Steel Quenching Degree

Abstract: The calculation of steel quenching degree has an important influence on real application. Steel quenching degree is influenced by chemical constitution and other many factors, which makes it difficult to be calculated accurately. Support vector machine (SVM) is a novel machine learning method based on statistical learning theory, which is powerful for solving the problems described by high dimension, small-sample and nonlinearity. In this paper, an SVM-based approach applied to calculate steel quenching degree is presented. With real data collected from Jiangyin Xingcheng Steel Work CO., LTD, experiments show that SVM-based method is effective and superior to ANN-based method.

Convex perceptrons

Abstract: Statistical learning theory make large margins an important property of linear classifiers and Support Vector Machines were designed with this target in mind. However, it has been shown that large margins can also be obtained when much simpler kernel perceptrons are used together with ad–hoc updating rules, different in principle from Rosenblatt’s rule. In this work we will numerically demonstrate that, rewritten in a convex update setting and using an appropriate updating vector selection procedure, Rosenblatt’s rule does indeed provide maximum margins for kernel perceptrons, although with a convergence slower than that achieved by other more sophisticated methods, such as the Schlesinger–Kozinec (SK) algorithm.