https://docs.google.com/presentation/d/1Jj3fOZFh_aZ2H4IOUcZts1OanU3bBMaN3zZDLEyRnno/edit#slide=id.g30800de2824_0_162

Perceptron: takes in three inputs each with a weight, and outputs a binary decision

• Activation Function

Loss Function: measures how well a given model fits the data

• lower the loss value = better the model fits the data

Optimizer: changes the weight of the neural network to reduce the error

CNN (Convolutional Neural Network)

1. Gathering Data

• Classification = one dimensional float
• Regression = 2D 3D values
  • if doing hand: 21 track points with XY values (21*2 = 42 inputs)

1. Preparing that Data
   • Hand: make a new array then just push xy values into it so it has 42 values
   • normalize or NOT