https://www.youtube.com/watch?v=CqOfi41LfDw
<aside> 💡
A neural network consists of Nodes and connections between the Nodes.
The numbers along each connection correspond to the parameter value estimated when this NN was fit to the data.
Think of them as slop and intercept considered when fitting a straight line to data
</aside>
The unknown parameter values in a NN is estimated when we fit to a dataset using a method called Backpropagation.
Nodes can have squiggles
These curves correspond to what method we use such as ReLU, SoftPlus or as you might know them by, Activation Function.
These are weights, we multiply them
These are biases, we add them
These are activation functions