基于复杂网络上监督学习方法的研究综述

A Review of Supervised Learning Methods Based on Complex Networks

复杂网络已经成为复杂系统分析问题的通用方法,随着人工智能和机器学习的广泛兴起,越来越多的学者开始关注在复杂网络上进行机器学习。监督学习作为机器学习的一个重要组成部分,本文深入研究和总结了基于复杂网络的监督学习方法。首先,本文分别从复杂网络和监督学习的理论基础入手,明确了相似性函数和相异性函数的概念和测度方法,系统梳理了复杂网络的构建方法,并阐明了监督学习的概念及其在机器学习中的地位。其次,介绍了监督学习的几种常用算法,梳理了各种算法的研究现状。然后,提出了基于复杂网络监督学习方法未来关注方向。最后,说明了基于复杂网络监督学习方法的局限性,为相关学者的研究提供了参考。

Complex networks have become a general method for complex system analysis. With the wide rise of artificial intelligence and machine learning, more and more scholars are beginning to pay attention to machine learning on complex networks. As an important part of machine learning, supervised learning is studied and summarized in this paper. Firstly, starting from the theoretical basis of complex networks and supervised learning, this paper clarifies the concepts and measurement methods of similarity function and anisotropy function, systematically sorts out the construction methods of complex networks, and expounds the concept of supervised learning and its position in machine learning. Secondly, it introduces several commonly used algorithms of supervised learning and sorts out the research status of various algorithms. Then, the future direction of supervised learning based on complex networks is proposed. Finally, the limitations of supervised learning methods based on complex networks are explained, which provides references for relevant scholars.