Neural Networks

Refer to Introduction to Neural Networks

[Personal Code Example](https://github.com/lauradang/data-science-training/blob/master/neuron.py)

Neurons

General Process:

1. Takes input

2. Does some math

3. Outputs something

Weights (Aka Parameters)

• Connects each neurons in one layer to the neurons in the next layer

Activation Function

• Turns unbounded input into output that is bounded and thus, predictable

• eg. Sigmoid function

Sigmoid Function

$\frac{1}{1+e^-x}$ $Range: [0, 1]$

• eg. Binary classification: The output would either be 0 or 1

What is happening in step 2 of process

1. Each input multiplied by some weight

$x_1$ → $x_1 * w_1$

1. All weighted inputs summed (basically dot product of input and weights) and bias is added

($x_1 * w_1$) + ($x_2 * w_2$) + $b$

1. Sum is passed into activation function

Neural Network

• Neurons connected together

Hidden Layer

• Layers between input and output layers