17/11/03 MWhite's learning notes

---

1. Classification

Linear Regression :y∈R
Classification(logistic regression) : y∈{0,1,...}

1.1 Hypothesis Representation

1.2 Logistic regression's Cost Function

Simplified Cost Function

Simplified Cost Function

1.3 Logistic regression's Gradient Descent

Vectorized implementation:

1.4 Advanced Optimization

library function——fminunc()

function [jVal, gradient] = costFunction(theta)
  jVal = [...code to compute J(theta)...];
  gradient = [...code to compute derivative of J(theta)...];
end

options = optimset('GradObj', 'on', 'MaxIter', 100);
initialTheta = zeros(2,1);
   [optTheta, functionVal, exitFlag] = fminunc(@costFunction, initialTheta, options);

2. Multiclass Classification

One-vs-all

3. Overfitting

skips θ0

3.1 Regularized Linear Regression

• Cost Function

  
  

• Gradient descent

  
  

• Normal Equation

  
  

3.2 Regularized Logistic Regression

• Cost Function

  
  

• Gradient descent