文件名称：基于支持向量机的机器学习研究 Research of Machine-Learning Based Support Vector Machine
文件大小：578KB
文件格式：CAJ
更新时间：2023-07-25 09:50:44
机器学习； 统计学习理论 支持向量机 SMO 人脸检测
【摘要】 学习是一切智能系统最根本的特征。机器学习是人工智能最具智能特征、最前沿的研究领域之一。机器学习研究的是如何使机器通过识别和利用现有知识来获取新知识和新技能。机器学习就是要使计算机能模拟人的学习行为,自动地通过学习获取知识和技能,不断改善性能,实现自我完善。 与传统统计学相比,统计学习理论是一种专门研究小样本情况下机器学习规律的理论。V.Vapnik 等人从六、七十年代开始致力于此方面研究,到九十年代中期,其理论不断发展和成熟。统计学习理论是建立在一套较坚实的理论基础之上的,为解决有限样本学习问题提供了一个统一的框架,它能将很多现有方法纳入其中,同时,在这一理论基础上发展了一种新的通用学习方法——支持向量机(Support Vector Machine 或 SVM),它已初步表现出很多优于已有方法的性能。 本文对机器学习、支持向量机的研究现状及应用领域进行了综述,阐述了机器学习和支持向量机的基本概念、基本模型和支持向量机的训练算法。针对机器学习系统的具体结构,提出了机器学习系统的模块化设计,划分出了输入处理、训练、执行与评价、评价表示 4 个模块,设计了各个模块之间的通信方式,并具体实现了 4 个模块和模块集成系统。 根据基于支持向量机的机器学习的研究成果,研制开发出人脸检测系统,主要包括人脸图像处理和编码、基于支持向量机的机器学习、执行与评价、评价表示功能,实现了人脸的自动判定。 还原 【Abstract】 Learning is the fundamental feather of all intelligent system.machine-learning is a domain with most intelligent feather and a domain with most foreland.machine-learning research how to obtain new knowledge and new technique by recognition and using existing knowledge.machine-learning let computer simulate man’s learning-action,automatically obtain new knowledge and new technique by learn,improve technique,realize own-perfect. To traditional statistics,Statistical Learning Theory is a theory of researching rule of machine-learning under little sample number.From 1960’s,V.Vapnik etc. begin to research this theory.To middle of 1990’s,this theory increasivly improved and completed.Statistical Learning Theory build on a set of stabile theory,offer a united framework to resolve little sample learning question,it can hold lots of existing method,and it build a new currency learning method--Support Vector Machine,Support Vector Machine method had represent better performance than existing method. This article summarize today state and application domain of machine-learning and Support Vector Machine,expound basic concept and basic model of machine-learning and Support Vector Machine and training arithmetic of Support Vector Machine.Aim at structure of machine-learning system,this article put forward module-design of machine-learning system,partition four module:deal-input,train,evaluate and evaluate representation,and design communicate style of each module,realize this four module and module-integration system. On result of researching machine-learning based Support Vector Machine,Face-Detection system is producted.This system mainly include dealing image of man-face and encode,machine-learning based Support Vector Machine,evaluate and evaluate representation,realize automatically judging man-face. 还原