• 【udacity】机器学习-神经网络


    Evernote Export

    1.神经网络

    神经元

    细胞的主体称为细胞体,然后有轴突、突触
    他们构建的方式是可以调整的
    我们会有一些输入的放电信号视为放电频率或输入的强度
    X1w1X2w2X3w3
    权值w基本相当于神经元对每个输入的增益或敏感性
    计算这些输入的和i=1kxiwiθ
    输出的结果如果是1或0,我们称这样的模型为神经网络中的感知器。

    2.感知器函数

    阶跃函数

    在数学中,如果实数域上的某个函数可以用半开区间上的指示函数的有限次线性组合来表示,那么这个函数就是阶跃函数,或者叫赫维赛德函数。换一种不太正式的说法就是,阶跃函数是有限段分段常数函数的组合。
    感知器在目前的情况下就是一系列的输入,返回值为0或1

    8.感知器训练

    1.感知器规则-->感知器使用的是阈值机制
    2.梯度下降(德尔塔定律)-->不使用阈值机制
    感知器最终只要得到的是一个线性的函数并且最终能够实现分类就能完成任务

    9.梯度下降

    10.学习规则比较

    1.对于感知器来说,线性的变化更适合,因为这是一个线性的规则,具有更好的收敛性
    2.对于梯度下降来说,更适合非线性的操作,因为使用的是微积分导数

    11.sigmod函数

    σ(a)=1+ea1
    a>σ(a)>0
    a>+σ(a)>1
    sigmod函数的特性就是随着激活函数趋向于负无穷,函数值趋向于0,随着激活函数趋向于正无穷,函数值趋向于1,并且整个过程是平滑曲线
    sigmod函数

    13.神经网络草图

    反向传播,从输入到输出,再回到输入来验证结果

    15.限定偏差

    限定偏差所用的任意一张数据结构,也可以告诉你将要考虑的假设集合
    感知器规则一定是线性的,sigmod函数结合起来可以表示更多问题
    我们训练神经网络的时候,一般会指定有限数量的隐藏单元以及有限数量的分层
    因此,并不是任何的神经网络真的都可以得到任意函数,任何固定的网络只能得到可以得到的结果
    网络的复杂性不仅仅由节点和分层决定,也和权值的大小有关

    16.喜好偏差

    选择偏置:我们会优先选择节点靠近根节点,有较高信息增益的树,我们优先选择正确的树

    17.小结

    1.感知器:一个阈值单元或者线性阈值单元
    2.感知器网络连接构成网络
    3.感知器的学习规则->针对学习规则

    %23%23%23%23%201.%E7%A5%9E%E7%BB%8F%E7%BD%91%E7%BB%9C%0A**%E7%A5%9E%E7%BB%8F%E5%85%83**%0A!%5B1f4de4d9c8aa96fd9a3f931d13c28842.png%5D(en-resource%3A%2F%2Fdatabase%2F1236%3A1)%0A%E7%BB%86%E8%83%9E%E7%9A%84%E4%B8%BB%E4%BD%93%E7%A7%B0%E4%B8%BA%E7%BB%86%E8%83%9E%E4%BD%93%EF%BC%8C%E7%84%B6%E5%90%8E%E6%9C%89%E8%BD%B4%E7%AA%81%E3%80%81%E7%AA%81%E8%A7%A6%0A%E4%BB%96%E4%BB%AC%E6%9E%84%E5%BB%BA%E7%9A%84%E6%96%B9%E5%BC%8F%E6%98%AF%E5%8F%AF%E4%BB%A5%E8%B0%83%E6%95%B4%E7%9A%84%0A%E6%88%91%E4%BB%AC%E4%BC%9A%E6%9C%89%E4%B8%80%E4%BA%9B%E8%BE%93%E5%85%A5%E7%9A%84%E6%94%BE%E7%94%B5%E4%BF%A1%E5%8F%B7%E8%A7%86%E4%B8%BA%E6%94%BE%E7%94%B5%E9%A2%91%E7%8E%87%E6%88%96%E8%BE%93%E5%85%A5%E7%9A%84%E5%BC%BA%E5%BA%A6%0A%24X_1w_1%24%24X_2w_2%24%24X_3w_3%24%0A%E6%9D%83%E5%80%BC%24w%24%E5%9F%BA%E6%9C%AC%E7%9B%B8%E5%BD%93%E4%BA%8E%E7%A5%9E%E7%BB%8F%E5%85%83%E5%AF%B9%E6%AF%8F%E4%B8%AA%E8%BE%93%E5%85%A5%E7%9A%84%E5%A2%9E%E7%9B%8A%E6%88%96%E6%95%8F%E6%84%9F%E6%80%A7%0A%E8%AE%A1%E7%AE%97%E8%BF%99%E4%BA%9B%E8%BE%93%E5%85%A5%E7%9A%84%E5%92%8C%24%5Csum%5Ek_%7Bi%3D1%7Dx_iw_i%5Cgeq%5Ctheta%24%0A%E8%BE%93%E5%87%BA%E7%9A%84%E7%BB%93%E6%9E%9C%E5%A6%82%E6%9E%9C%E6%98%AF1%E6%88%960%EF%BC%8C%E6%88%91%E4%BB%AC%E7%A7%B0%E8%BF%99%E6%A0%B7%E7%9A%84%E6%A8%A1%E5%9E%8B%E4%B8%BA%E7%A5%9E%E7%BB%8F%E7%BD%91%E7%BB%9C%E4%B8%AD%E7%9A%84%E6%84%9F%E7%9F%A5%E5%99%A8%E3%80%82%0A%0A%23%23%23%23%202.%E6%84%9F%E7%9F%A5%E5%99%A8%E5%87%BD%E6%95%B0%0A**%5B%E9%98%B6%E8%B7%83%E5%87%BD%E6%95%B0%5D(https%3A%2F%2Fzh.wikipedia.org%2Fzh-cn%2F%25E9%2598%25B6%25E8%25B7%2583%25E5%2587%25BD%25E6%2595%25B0)**%0A%3E%E5%9C%A8%E6%95%B0%E5%AD%A6%E4%B8%AD%EF%BC%8C%E5%A6%82%E6%9E%9C%E5%AE%9E%E6%95%B0%E5%9F%9F%E4%B8%8A%E7%9A%84%E6%9F%90%E4%B8%AA%E5%87%BD%E6%95%B0%E5%8F%AF%E4%BB%A5%E7%94%A8%E5%8D%8A%E5%BC%80%E5%8C%BA%E9%97%B4%E4%B8%8A%E7%9A%84%E6%8C%87%E7%A4%BA%E5%87%BD%E6%95%B0%E7%9A%84%E6%9C%89%E9%99%90%E6%AC%A1%E7%BA%BF%E6%80%A7%E7%BB%84%E5%90%88%E6%9D%A5%E8%A1%A8%E7%A4%BA%EF%BC%8C%E9%82%A3%E4%B9%88%E8%BF%99%E4%B8%AA%E5%87%BD%E6%95%B0%E5%B0%B1%E6%98%AF%E9%98%B6%E8%B7%83%E5%87%BD%E6%95%B0%EF%BC%8C%E6%88%96%E8%80%85%E5%8F%AB%E8%B5%AB%E7%BB%B4%E8%B5%9B%E5%BE%B7%E5%87%BD%E6%95%B0%E3%80%82%E6%8D%A2%E4%B8%80%E7%A7%8D%E4%B8%8D%E5%A4%AA%E6%AD%A3%E5%BC%8F%E7%9A%84%E8%AF%B4%E6%B3%95%E5%B0%B1%E6%98%AF%EF%BC%8C%E9%98%B6%E8%B7%83%E5%87%BD%E6%95%B0%E6%98%AF%E6%9C%89%E9%99%90%E6%AE%B5%E5%88%86%E6%AE%B5%E5%B8%B8%E6%95%B0%E5%87%BD%E6%95%B0%E7%9A%84%E7%BB%84%E5%90%88%E3%80%82%0A%E6%84%9F%E7%9F%A5%E5%99%A8%E5%9C%A8%E7%9B%AE%E5%89%8D%E7%9A%84%E6%83%85%E5%86%B5%E4%B8%8B%E5%B0%B1%E6%98%AF%E4%B8%80%E7%B3%BB%E5%88%97%E7%9A%84%E8%BE%93%E5%85%A5%EF%BC%8C%E8%BF%94%E5%9B%9E%E5%80%BC%E4%B8%BA0%E6%88%961%0A%0A%23%23%23%23%208.%E6%84%9F%E7%9F%A5%E5%99%A8%E8%AE%AD%E7%BB%83%0A1.%E6%84%9F%E7%9F%A5%E5%99%A8%E8%A7%84%E5%88%99--%3E%E6%84%9F%E7%9F%A5%E5%99%A8%E4%BD%BF%E7%94%A8%E7%9A%84%E6%98%AF%E9%98%88%E5%80%BC%E6%9C%BA%E5%88%B6%0A2.%E6%A2%AF%E5%BA%A6%E4%B8%8B%E9%99%8D(%E5%BE%B7%E5%B0%94%E5%A1%94%E5%AE%9A%E5%BE%8B)--%3E%E4%B8%8D%E4%BD%BF%E7%94%A8%E9%98%88%E5%80%BC%E6%9C%BA%E5%88%B6%0A%E6%84%9F%E7%9F%A5%E5%99%A8%E6%9C%80%E7%BB%88%E5%8F%AA%E8%A6%81%E5%BE%97%E5%88%B0%E7%9A%84%E6%98%AF%E4%B8%80%E4%B8%AA%E7%BA%BF%E6%80%A7%E7%9A%84%E5%87%BD%E6%95%B0%E5%B9%B6%E4%B8%94%E6%9C%80%E7%BB%88%E8%83%BD%E5%A4%9F%E5%AE%9E%E7%8E%B0%E5%88%86%E7%B1%BB%E5%B0%B1%E8%83%BD%E5%AE%8C%E6%88%90%E4%BB%BB%E5%8A%A1%0A%0A%23%23%23%23%209.%E6%A2%AF%E5%BA%A6%E4%B8%8B%E9%99%8D%0A%0A%23%23%23%23%2010.%E5%AD%A6%E4%B9%A0%E8%A7%84%E5%88%99%E6%AF%94%E8%BE%83%0A1.%E5%AF%B9%E4%BA%8E%E6%84%9F%E7%9F%A5%E5%99%A8%E6%9D%A5%E8%AF%B4%EF%BC%8C%E7%BA%BF%E6%80%A7%E7%9A%84%E5%8F%98%E5%8C%96%E6%9B%B4%E9%80%82%E5%90%88%EF%BC%8C%E5%9B%A0%E4%B8%BA%E8%BF%99%E6%98%AF%E4%B8%80%E4%B8%AA%E7%BA%BF%E6%80%A7%E7%9A%84%E8%A7%84%E5%88%99%EF%BC%8C%E5%85%B7%E6%9C%89%E6%9B%B4%E5%A5%BD%E7%9A%84%E6%94%B6%E6%95%9B%E6%80%A7%0A2.%E5%AF%B9%E4%BA%8E%E6%A2%AF%E5%BA%A6%E4%B8%8B%E9%99%8D%E6%9D%A5%E8%AF%B4%EF%BC%8C%E6%9B%B4%E9%80%82%E5%90%88%E9%9D%9E%E7%BA%BF%E6%80%A7%E7%9A%84%E6%93%8D%E4%BD%9C%EF%BC%8C%E5%9B%A0%E4%B8%BA%E4%BD%BF%E7%94%A8%E7%9A%84%E6%98%AF%E5%BE%AE%E7%A7%AF%E5%88%86%E5%AF%BC%E6%95%B0%0A%0A%23%23%23%23%2011.sigmod%E5%87%BD%E6%95%B0%0A%24%5Csigma(a)%3D%5Cfrac%7B1%7D%7B1%2Be%5E%7B-a%7D%7D%24%0A%24a-%3E-%5Cinfty%20%5Csigma(a)-%3E0%24%0A%24a-%3E%2B%5Cinfty%20%5Csigma(a)-%3E1%24%0Asigmod%E5%87%BD%E6%95%B0%E7%9A%84%E7%89%B9%E6%80%A7%E5%B0%B1%E6%98%AF%E9%9A%8F%E7%9D%80%E6%BF%80%E6%B4%BB%E5%87%BD%E6%95%B0%E8%B6%8B%E5%90%91%E4%BA%8E%E8%B4%9F%E6%97%A0%E7%A9%B7%EF%BC%8C%E5%87%BD%E6%95%B0%E5%80%BC%E8%B6%8B%E5%90%91%E4%BA%8E0%EF%BC%8C%E9%9A%8F%E7%9D%80%E6%BF%80%E6%B4%BB%E5%87%BD%E6%95%B0%E8%B6%8B%E5%90%91%E4%BA%8E%E6%AD%A3%E6%97%A0%E7%A9%B7%EF%BC%8C%E5%87%BD%E6%95%B0%E5%80%BC%E8%B6%8B%E5%90%91%E4%BA%8E1%EF%BC%8C%E5%B9%B6%E4%B8%94%E6%95%B4%E4%B8%AA%E8%BF%87%E7%A8%8B%E6%98%AF%E5%B9%B3%E6%BB%91%E6%9B%B2%E7%BA%BF%0A**%5Bsigmod%E5%87%BD%E6%95%B0%5D(http%3A%2F%2Fmathworld.wolfram.com%2FSigmoidFunction.html)**%0A%0A%23%23%23%23%2013.%E7%A5%9E%E7%BB%8F%E7%BD%91%E7%BB%9C%E8%8D%89%E5%9B%BE%0A%E5%8F%8D%E5%90%91%E4%BC%A0%E6%92%AD%EF%BC%8C%E4%BB%8E%E8%BE%93%E5%85%A5%E5%88%B0%E8%BE%93%E5%87%BA%EF%BC%8C%E5%86%8D%E5%9B%9E%E5%88%B0%E8%BE%93%E5%85%A5%E6%9D%A5%E9%AA%8C%E8%AF%81%E7%BB%93%E6%9E%9C%0A%0A%23%23%23%23%2015.%E9%99%90%E5%AE%9A%E5%81%8F%E5%B7%AE%0A%3E%E9%99%90%E5%AE%9A%E5%81%8F%E5%B7%AE%E6%89%80%E7%94%A8%E7%9A%84%E4%BB%BB%E6%84%8F%E4%B8%80%E5%BC%A0%E6%95%B0%E6%8D%AE%E7%BB%93%E6%9E%84%EF%BC%8C%E4%B9%9F%E5%8F%AF%E4%BB%A5%E5%91%8A%E8%AF%89%E4%BD%A0%E5%B0%86%E8%A6%81%E8%80%83%E8%99%91%E7%9A%84%E5%81%87%E8%AE%BE%E9%9B%86%E5%90%88%0A%E6%84%9F%E7%9F%A5%E5%99%A8%E8%A7%84%E5%88%99%E4%B8%80%E5%AE%9A%E6%98%AF%E7%BA%BF%E6%80%A7%E7%9A%84%EF%BC%8Csigmod%E5%87%BD%E6%95%B0%E7%BB%93%E5%90%88%E8%B5%B7%E6%9D%A5%E5%8F%AF%E4%BB%A5%E8%A1%A8%E7%A4%BA%E6%9B%B4%E5%A4%9A%E9%97%AE%E9%A2%98%0A%E6%88%91%E4%BB%AC%E8%AE%AD%E7%BB%83%E7%A5%9E%E7%BB%8F%E7%BD%91%E7%BB%9C%E7%9A%84%E6%97%B6%E5%80%99%EF%BC%8C%E4%B8%80%E8%88%AC%E4%BC%9A%E6%8C%87%E5%AE%9A%E6%9C%89%E9%99%90%E6%95%B0%E9%87%8F%E7%9A%84%E9%9A%90%E8%97%8F%E5%8D%95%E5%85%83%E4%BB%A5%E5%8F%8A%E6%9C%89%E9%99%90%E6%95%B0%E9%87%8F%E7%9A%84%E5%88%86%E5%B1%82%0A%E5%9B%A0%E6%AD%A4%EF%BC%8C%E5%B9%B6%E4%B8%8D%E6%98%AF%E4%BB%BB%E4%BD%95%E7%9A%84%E7%A5%9E%E7%BB%8F%E7%BD%91%E7%BB%9C%E7%9C%9F%E7%9A%84%E9%83%BD%E5%8F%AF%E4%BB%A5%E5%BE%97%E5%88%B0%E4%BB%BB%E6%84%8F%E5%87%BD%E6%95%B0%EF%BC%8C%E4%BB%BB%E4%BD%95%E5%9B%BA%E5%AE%9A%E7%9A%84%E7%BD%91%E7%BB%9C%E5%8F%AA%E8%83%BD%E5%BE%97%E5%88%B0%E5%8F%AF%E4%BB%A5%E5%BE%97%E5%88%B0%E7%9A%84%E7%BB%93%E6%9E%9C%0A**%E7%BD%91%E7%BB%9C%E7%9A%84%E5%A4%8D%E6%9D%82%E6%80%A7%E4%B8%8D%E4%BB%85%E4%BB%85%E7%94%B1%E8%8A%82%E7%82%B9%E5%92%8C%E5%88%86%E5%B1%82%E5%86%B3%E5%AE%9A%EF%BC%8C%E4%B9%9F%E5%92%8C%E6%9D%83%E5%80%BC%E7%9A%84%E5%A4%A7%E5%B0%8F%E6%9C%89%E5%85%B3**%0A%0A%23%23%23%23%2016.%E5%96%9C%E5%A5%BD%E5%81%8F%E5%B7%AE%0A%E9%80%89%E6%8B%A9%E5%81%8F%E7%BD%AE%EF%BC%9A%E6%88%91%E4%BB%AC%E4%BC%9A%E4%BC%98%E5%85%88%E9%80%89%E6%8B%A9%E8%8A%82%E7%82%B9%E9%9D%A0%E8%BF%91%E6%A0%B9%E8%8A%82%E7%82%B9%EF%BC%8C%E6%9C%89%E8%BE%83%E9%AB%98%E4%BF%A1%E6%81%AF%E5%A2%9E%E7%9B%8A%E7%9A%84%E6%A0%91%EF%BC%8C%E6%88%91%E4%BB%AC%E4%BC%98%E5%85%88%E9%80%89%E6%8B%A9%E6%AD%A3%E7%A1%AE%E7%9A%84%E6%A0%91%0A%0A%23%23%23%23%2017.%E5%B0%8F%E7%BB%93%0A1.%E6%84%9F%E7%9F%A5%E5%99%A8%EF%BC%9A%E4%B8%80%E4%B8%AA%E9%98%88%E5%80%BC%E5%8D%95%E5%85%83%E6%88%96%E8%80%85%E7%BA%BF%E6%80%A7%E9%98%88%E5%80%BC%E5%8D%95%E5%85%83%0A2.%E6%84%9F%E7%9F%A5%E5%99%A8%E7%BD%91%E7%BB%9C%E8%BF%9E%E6%8E%A5%E6%9E%84%E6%88%90%E7%BD%91%E7%BB%9C%0A3.%E6%84%9F%E7%9F%A5%E5%99%A8%E7%9A%84%E5%AD%A6%E4%B9%A0%E8%A7%84%E5%88%99-%3E%E9%92%88%E5%AF%B9%E5%AD%A6%E4%B9%A0%E8%A7%84%E5%88%99

    Win a contest, win a challenge
  • 相关阅读:
    django1.6 django-dajaxice的安装配置.
    jquery js ajax 不错的想法
    django1.6 GET url传参 乱码
    django1.6 CSRF verification failed. Request aborted. 用出现表单提交
    Caml语句中筛选lookup字段
    sharepoint Jsom一些基本操作
    JSOM启动工作流
    数据量太大时,如何实现分页查询-CSOM
    数据量太大时,如何实现分页查询-JSOM
    JS加强学习-DOM学习02
  • 原文地址:https://www.cnblogs.com/pandaboy1123/p/10253940.html
Copyright © 2020-2023  润新知