码迷,mamicode.com
首页 > 其他好文 > 详细

什么是机器学习

时间:2018-10-15 18:30:58      阅读:154      评论:0      收藏:0      [点我收藏+]

标签:学习   基于   label   垃圾   ORC   记录   rac   style   under   

机器学习的定义    

  • 定义1
    • Arthur Samuel (1959). Machine Learning: Field of study that gives computers the ability to learn without being explicitly programmed.
    • Arthur Samuel 在1959 年给出的定义:机器学习是一种在不明确编程的情况下使计算机具有学习能力的研究领域,
  • 定义2 (不太容易理解)
    • Tom Mitchell (1998) Well-posed Learning Problem: A computer program is said to learn from experience E with respect to some task T and some performance measure P, if its performance on T, as measure by P, improves with experience E.
    • Tom Mitchell 在 1998 年给出的:(谷歌翻译)适当的学习问题:是计算机程序从经验 E 中学习某个任务 T 和一些绩效测量 P,如果它在 T 上的表现,由 P 测量,随经验 E 而改善。(稍后根据例子理解)
    • #################################################################
    • Example : suppose your email program watches which emails you do or do not mark as spam, and based on that learns how to better filter spam. What is the task T in this setting?
      • Classifying emails as spam or not spam.  --- T
      • Watching you label emails as spam or not spam.  --- E
      • The number (or fraction) of emails correctly classified as spam/not spam. --- P
    • 例子:假设你的 E-MAIL 程序观察你标记的垃圾邮件和非垃圾邮件后,基于观察结果学会了怎样才能更好的过滤垃圾邮件,这个例子中的 T 是什么?
      • 将邮件分类为垃圾邮件和非垃圾邮件  --- T
      • 看着你将电子邮件标记为垃圾邮件或非垃圾邮件  ---E
      • 正确分类为垃圾邮件/非垃圾邮件的电子邮件数量(或分数) ---P
    • Example: playing checkers.
      • E = the experience of playing many games of checkers

      • T = the task of playing checkers.

      • P = the probability that the program will win the next game.
    • 例子:跳棋游戏
      • E = 玩很多距跳棋的经历
      • T = 玩跳棋的任务
      • P = 程序赢下游戏的可能性

机器学习的分类

  • Supervised learning (监督学习)
  • Unsupervised learning (非监督学习)
  • others: Reinforcement learning, recommender systems. (增强学习、推荐系统)

为什么现在这么受欢迎

    Machine learning

  • Grew out of work in AI (从传统人工智能中脱离出来(可以理解为有自己的一套体系))
  • New capability for computers (为计算机提供了一种新的能力(计算机可以自己学习))

机器应用领域

  Examples:

  • Database mining (数据挖掘)
    • large datasets from growth of automation/web. (自动化/网络增长的大型数据集)
    • E.g, Web click data, medical records, biology, engineering. (网络点击数据、医学记录等)
  • Applications can‘t program by hand (无法手动编程的应用)
    • E.g, Autonomous helicopter, handwriting recognition, most of Natural Language Processing (NLP), Computer Vision. (自主直升机、手写识别、自然语言处理以及计算机视觉等)
  • Self-customizing programs (定制化程序)
    • E.g, Amazon, Netfix product recommendations (产品推荐(比如:按照喜好或购买记录推荐商品))
  • Understanding human learning(brain, real AI). (机器学习就是明白人类是如何学习的)

什么是机器学习

标签:学习   基于   label   垃圾   ORC   记录   rac   style   under   

原文地址:https://www.cnblogs.com/qkloveslife/p/9791204.html

(0)
(0)
   
举报
评论 一句话评论(0
登录后才能评论!
© 2014 mamicode.com 版权所有  联系我们:gaon5@hotmail.com
迷上了代码!