Introduction to Neural Networks

Neural networks are a more sophisticated version of feature crosses. In essence, neural networks learn the appropriate feature crosses for you.

Intro to Neural Networks

A Linear Model

Three blue circles in a row connected by arrows to a green circle above them image/svg+xml Input Output

Add Complexity: Non-Linear?

Three blue circles in a row labeled "Input" connected by arrows to a row of yellow circles labeled "Hidden Layer" above them, which are in turn connected to a green circle labeled "Output" at the top. image/svg+xml Output Hidden Layer Input

More Complex: Non-Linear?

image/svg+xml Output Hidden Layer 2 Hidden Layer 1 Input

Adding a Non-Linearity

The same as the previous figure, except that a row of pink circles labeled 'Non-Linear Transformation Layer' has been added in between the two hidden layers. image/svg+xml Output Hidden Layer 2 Non-Linear Transformation Layer (a.k.a. Activation Function) Hidden Layer 1 Input We Usually Don't Draw Non-Linear Transforms

Our Favorite Non-Linearity

A graph with slope of 0 and then linear once it passes x=0 image/svg+xml Relu Rectified Linear Unit F(x)=max(0,x)

Neural Nets Can Be Arbitrarily Complex

A complex neural network image/svg+xml Hidden2 Hidden1 Input Output

Send feedback about...

Machine Learning Crash Course