机器学习（1）——概览

一、机器学习是什么？

  Machine learning  is the subfield of  computer science  that gives computers the ability to learn without being explicitly programmed ( Arthur Samuel , 1959). ^[1]  Evolved from the study of  pattern recognition  and  computational learning theory  in  artificial intelligence , ^[2]  machine learning explores the study and construction of  algorithms  that can  learn  from and make predictions on  data ^[3]  – such algorithms overcome following strictly static  program instructions  by making data driven predictions or decisions, ^{[4] :2}  through building a  model  from sample inputs. Machine learning is employed in a range of computing tasks where designing and programming explicit  algorithms  is infeasible; example applications include  spam filtering , detection of network intruders or malicious insiders working towards a data breach, ^[5]   optical character recognition  (OCR), ^[6]   search engines  and  computer vision .

Machine learning is closely related to (and often overlaps with) computational statistics, which also focuses in prediction-making through the use of computers. It has strong ties to mathematical optimization, which delivers methods, theory and application domains to the field. Machine learning is sometimes conflated with data mining,^[7] where the latter subfield focuses more on exploratory data analysis and is known as unsupervised learning.^[4]:vii[8]Machine learning can also be unsupervised^[9] and be used to learn and establish baseline behavioral profiles for various entities^[10] and then used to find meaningful anomalies.

Within the field of data analytics, machine learning is a method used to devise complex models and algorithms that lend themselves to prediction; in commercial use, this is known as predictive analytics. These analytical models allow researchers, data scientists, engineers, and analysts to "produce reliable, repeatable decisions and results" and uncover "hidden insights" through learning from historical relationships and trends in the data.

机器学习算法是一类从数据中自动分析获得规律，并利用规律对未知数据进行预测的算法.

二、机器学习的过程。

 构建算法模型——获取数据——提取数据特征——训练模型——优化模型——预测分析

三、分类。

• 监督学习从给定的训练数据集中学习出一个函数，当新的数据到来时，可以根据这个函数预测结果。监督学习的训练集要求是包括输入和输出，也可以说是特征和目标。训练集中的目标是由人标注的。常见的监督学习算法包括回归分析和统计分类。
• 无监督学习与监督学习相比，训练集没有人为标注的结果。常见的无监督学习算法有聚类。
• 半监督学习介于监督学习与无监督学习之间。
• 增强学习通过观察来学习做成如何的动作。每个动作都会对环境有所影响，学习对象根据观察到的周围环境的反馈来做出判断。

四、算法模型。

五、主要应用

自适应网站
情感计算
生物信息学
脑机接口
化学信息学
分类DNA序列
计算解剖
计算机视觉，包括对象识别
检测信用卡欺诈
游戏
信息检索
网络欺诈检测
市场
机器感知
医疗诊断
经济
自然语言处理
自​​然语言理解
优化和元启发
在线广告
推荐系统
机器人移动
搜索引擎
情感分析（或意见挖掘）
序列挖掘
软件工程
语音和手写识别
股市分析
结构健康监测
句法模式识别
用户行为分析

参考资料：

1、https://en.wikipedia.org/wiki/Machine_learning

2、https://zh.wikipedia.org/wiki/%E6%9C%BA%E5%99%A8%E5%AD%A6%E4%B9%A0

3、机器学习实战   Peter Harrington

拓展阅读：

1、

了解点机器学习

2、

机器学习与数据挖掘 -国内牛人网站分享（免费）