Topic
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.
Papers published on a yearly basis
Papers
More filters
•
TL;DR: Experimental results show that the improved one-against-one support vector machine approach is a promising technique for image segmentation.
Abstract: Support vector machine approach is considered a good candidate because of its good generalization performance, especially when the number of training samples is very small and the dimension of feature space is very high. In this paper, an improved one-against-one support vector machine is proposed and the segmentation of multi-target image based on the improved one-against-one support vector machine approach is investigated. Experimental results show that support vector machine approach is a promising technique for image segmentation.
1 citations
•
TL;DR: An object classification algorithm is proposed in this paper used for video surveillance, which can classify moving objects into predefined four categories: human being, crowd, car and bicyclist.
Abstract: Video-based motion analysis is receiving increasing attention in the domain of computer vision.An object classification algorithm is proposed in this paper used for video surveillance,which can classify moving objects into predefined four categories: human being,crowd,car and bicyclist.In this paper,several simple shape features of moving objects are defined and the SVM(Support Vector Machines),which based on small samples statistical learning theory,is chosen to classify different objects.At last,to meet the real-time requirement,the method of alternative classification is presented and several other methods that can improve the efficiency of classification are described.Experiments show that the method can accurately distinguish human being,crowd,car,and bicyclist.
1 citations
••
18 Oct 2010TL;DR: KM is described, which is one of the most interesting results of statistical learning theory capable to abstract system design and make it simpler, and an example of effective use of KM for the design of a natural language application required in the European Project LivingKnowledge.
Abstract: Latest results of statistical learning theory have provided techniques such us pattern analysis and relational learning, which help in modeling system behavior, e.g. the semantics expressed in text, images, speech for information search applications (e.g. as carried out by Google, Yahoo,..) or the semantics encoded in DNA sequences studied in Bioinformatics. These represent distinguished cases of successful use of statistical machine learning. The reason of this success relies on the ability of the latter to overcome the critical limitations of logic/rule-based approaches to semantic modeling: although, from a knowledge engineer perspective, hand-crafted rules are natural methods to encode system semantics, noise, ambiguity and errors, affecting dynamic systems, prevent them from being effective.
One drawback of statistical approaches relates to the complexity of modeling world objects in terms of simple parameters. In this paper, we describe kernel methods (KM), which are one of the most interesting results of statistical learning theory capable to abstract system design and make it simpler. We provide an example of effective use of KM for the design of a natural language application required in the European Project LivingKnowledge.
1 citations
••
TL;DR: This paper adopts statistical learning theory and optimization theory to the analysis of the algorithm theory, probe into its theoretical foundation.
Abstract: This paper adopts statistical learning theory and optimization theory to the analysis of the algorithm theory, probe into its theoretical foundation. The existing theoretical analysis on the basis of the establishment of clustering model algorithm design, code realization and finally a lot of different data set of test, choose soil data as a test database, will be in the database on a large number of data mining experiment to verify the performance of the proposed algorithm. The test result feedback back will further deepen the theoretical research or correct theory already mistakes, new theory and will continue to guide experiments, both mutual promoting common development.
1 citations
••
16 Jun 2011
TL;DR: An intelligent distribution identification model was established based on statistical learning theory and the algorithm of multi-element classifier of Support Vector Machine and also applied to parameter estimation of small sample censored data, in order to improve the precision of traditional method.
Abstract: It is difficult to identify distribution types and to estimate parameters of the distribution for small sample censored data. An intelligent distribution identification model was established based on statistical learning theory and the algorithm of multi-element classifier of Support Vector Machine (SVM), and also applied to parameter estimation of small sample censored data, in order to improve the precision of traditional method. The algorithm of training based on SVM and the RBF kernel function was selected firstly; secondly, the parameters of the distributions characteristics were drawn; on the basis of these conditions, the distributions identification model and the parameter estimation model were constructed finally. The model was verified with Monte Carlo simulation method. Plenty of combinations of the numbers of training and testing data were processed to find the optimization of identification model and make it efficient. The results indicate that the new algorithm has more preferable performance in distribution type identification and parameter estimation than the traditional methods.
1 citations