Intro

• instances
• features
• labels
• classes
• regression
• classification
• overfitting

KNN

• neighbor
• distance
• kNN
• overfitting in kNN
• bias and variance in kNN

Software background

• objects vs classes
• methods
• locations of objects in memory
• instantiating an object from a class
• tuning a parameter and making a plot

Perceptron

• Perceptron decision rule
• learning, as in machine "learning"
• learnable parameter
• perceptron update rule
• bias term
• linear separability
• learning rate
• dot product

Learning and optimization

• grid search
• random search
• gradient
• gradient descent algorithm
• loss function

Logistic regression

• the logistic function
• likelihood function

Linear algebra background

• matrix vector operations
• matrix vector operations in numpy

Splitting a dataset

• train vs test set
• validation set
• cross validation

SVMs

• support vectors
• margins
• hinge loss

Multi-class classification

• one vs. rest
• one vs. all
• Softmax function