机器学习定义
Arthur Samuel (1959). Machine Learning: Field of study that gives computers the ability to learn without being explicitly programmed.
使机器在没有明确编程上的研究领域具备学习的能力.
![使机器在没有明确编程上的研究领域具备学习的能力](/2018/09/26/Coursera%E6%9C%BA%E5%99%A8%E5%AD%A6%E4%B9%A0%E5%90%B4%E6%81%A9%E8%BE%BE-Week1/1.png)
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 measured by P, improves with experience E.
对于某类任务T和性能度量P,如果一个计算机程序在T上以P衡量的性能随着经验E而自我完善,那么我们称这个计算机程序在从经验E学习.
完全为了押韵…
![定义2](/2018/09/26/Coursera%E6%9C%BA%E5%99%A8%E5%AD%A6%E4%B9%A0%E5%90%B4%E6%81%A9%E8%BE%BE-Week1/2.png)
机器学习算法
- 监督式学习
- 非监督式学习
- 强化学习
- 推荐系统
监督式学习介绍
![房屋价钱预测](/2018/09/26/Coursera%E6%9C%BA%E5%99%A8%E5%AD%A6%E4%B9%A0%E5%90%B4%E6%81%A9%E8%BE%BE-Week1/3.png)
如图片所示,监督式学习为right answers given.
.此处为regression(回归模型)
.
Regression: Predict continuous valued output(price)
回归: 预测一系列的值输出(价钱)
![乳腺癌(恶性,良性)](/2018/09/26/Coursera%E6%9C%BA%E5%99%A8%E5%AD%A6%E4%B9%A0%E5%90%B4%E6%81%A9%E8%BE%BE-Week1/4.png)
此处为Classification(分类模型)
,
Classification: Discrete valued output(0 or 1)
分类: 分离值输出(0或者1)
非监督式学习介绍
![监督式学习](/2018/09/26/Coursera%E6%9C%BA%E5%99%A8%E5%AD%A6%E4%B9%A0%E5%90%B4%E6%81%A9%E8%BE%BE-Week1/5.png)
![非监督式学习](/2018/09/26/Coursera%E6%9C%BA%E5%99%A8%E5%AD%A6%E4%B9%A0%E5%90%B4%E6%81%A9%E8%BE%BE-Week1/6.png)
模型与代价函数
代价函数: 与真实的所表现出来的误差函数.
- 线性回归: liner regression
- 损失函数: cost function
- 梯度下降: gradient descent
梯度下降
强烈建议阅读:深入浅出-梯度下降法及其实现
矩阵
略.