图核函数研究现状与进展

发布时间：2018-05-19 04:41

本文选题：结构化 + 图核　；参考：《安徽大学学报(自然科学版)》2017年01期

【摘要】：核方法具有坚实的理论基础和广泛的应用,已引起了各领域的关注.基于核的机器学习方法不仅适用于以特征向量表示的模式,也适用于结构化数据的模式.前者对应的是向量核方法,后者对应的是图核方法.图核对结构化数据具有强大而灵活的表示形式,其不仅能描述研究对象或模式的特性,还能反映构成这个物体不同部分之间的结构信息.目前,基于图核的机器学习方法在模式识别、机器学习、机器视觉、数据挖掘等相关研究领域得到了极为广泛的关注与应用,已成为结构数据描述方法和应用领域的一个重要研究方向.论文从使用最为广泛的基于R-convolution的图核谈起,总结了图核研究的意义,着重回顾和讨论图核函数的基本理论、基本分类、国内外研究现状,并进一步指出图核研究的发展方向.
[Abstract]:The nuclear method has a solid theoretical foundation and a wide range of applications, has attracted attention in various fields. The kernel-based machine learning method is not only applicable to the schema represented by eigenvector, but also applicable to the schema of structured data. The former corresponds to the vector kernel method, the latter corresponds to the graph kernel method. Graph checking structured data has a powerful and flexible representation, which can not only describe the characteristics of research object or pattern, but also reflect the structural information between the different parts of the object. At present, the machine learning method based on graph kernel has been widely concerned and applied in the fields of pattern recognition, machine learning, machine vision, data mining and so on. It has become an important research direction in the field of structural data description and application. This paper begins with the most widely used graph kernel based on R-convolution, summarizes the significance of graph kernel research, reviews and discusses the basic theory, basic classification, research status at home and abroad, and further points out the development direction of graph kernel research.
【作者单位】：中央财经大学信息学院;合肥学院数学与物理系;安徽大学计算机科学与技术学院;
【基金】：国家自然科学基金资助项目(61503422,61602535)
【分类号】：TP181

【相似文献】