ML 基础知识

标签：

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

ML Algorithms Overview

• Supervised learning    <= "teach" program
  • Given "right answers" data, then predict 
  • Regression: predict
• Unsupervisedlearning <= let it learn by itself
  • Given data without labels, then find some structures in the data
• Others: reinforcement learning, recommender systems 

Regression Overivew 

To get the prediction model, we need to define the hythontheis function, and determine the parameters

• Hythonthesis function & Cost Function
  
  • Hypothesis function hΘ(x)
  • Cost Function J(Θ)
• Gradient Descent

• Newton‘s method

Linear Regression

• Hypothesis function hΘ(x) = Θ^Tx 
• Gradient descent for linear regression

• Feature scaling
  • make sure features are on similar scales 
• Learning rate α 
  • pick the one seems to get J(Θ) to decrease fastest
• Features & Polynomial regession 
• Normal Equation 
  
  • too many features
    • regularization or delete some
    • redundent features (e.g. linear dependent features)

Logistic Regression 

• Hypothesis function:   [0,1]
• Gradient descent & Newton‘s method for logisitic regression

Regularization＊

Regularizatio（正则化）意在eliminate overfitting（过拟合）问题。因为参数太多，会导致我们的模型复杂度上升，容易过拟合，也就是我们的训练误差会很小。但训练误差小并不是我们的最终目标，我们的目标是希望模型的测试误差小，也就是能准确的预测新的样本。所以，我们需要保证模型“简单”的基础上最小化训练误差，这样得到的参数才具有好的泛化性能（也就是测试误差也小），而模型“简单”就是通过规则函数来实现的。

简单来说，我们需要在训练误差小（目标1）和模型简单（目标2）之间tradeoff! 

• 过拟合问题 （too many features）

• Regularized linear regression

• Regularized logistic regression

• regularization 惩罚项 & L2范数*

Reference

• http://www.52ml.net/12019.html
• http://blog.csdn.net/zouxy09/article/details/24971995/

ML 基础知识

标签：

原文地址：http://www.cnblogs.com/qingwen/p/5006969.html