Supervised Learning

Finding the right model to fit data

Like reverse Algebra: 

we only know x and y 

Try to find the min delta

Multiple Params

Find the best model

Classification

True of False

y = 1/(1+ e^(-theta * X))
so y will always be (0,1)

Identify more than one:
one vs all

Regression
FTW!!!!!

Overfitting

One vs all is not very good...

Picking param is hard :(

Neural Network FTW!!!

No time, next time!

Rise of cheap machine!