Neural Network Neckties

For when you want to signal to everyone that you've got a handle on the basics of Deep Learning

Gradient Descent

Because who needs a fancy algorithm when you can just slide into the minimum?

Stochastic Diffusion

Backpropagation

When you want to send errors flying in the opposite direction

Regularization

Autoencoders

Because who needs input when you can just repeat yourself?

Variational Bayes