Learning Basics
A kind of learning where our set of examples contain the right answer, also known as labels.
Supervised Learning
The likelihood of an event
Probability
The answers for every example.
Labels
Part 1 of the learning process where learning happens.
The training phase
A type of learning task where we try to predict a yes/no.
Binary Classification
The quantity produced by a random process; it represents a quantity whose value we are uncertain about.
Random Variable
The characteristics of a dataset; the columns of our examples.
Features
Part 2 of the learning process where we evaluate how well our QuAM performs
The Testing Phase
A type of learning task where we have a set of three or more options to predict.
Multi-class Classification
The probability of every possible state of a random variable.
Probability Distribution
A set of examples used to train a learning algorithm
Training Data
The output of a Learning Algorithm, a model of how to do something.
The QuAM!
The problem of predicting a real number.
Regression
The probabilities of all possible values of a random variable must add up to 1.
Normalisation Constraint
The set of examples that the QuAM uses to evaluate its performance.
Test Data
The output of a built QuAM which is generated after we input new (unseen) data.
The answer or predicted label
The goal of ML.
"Find the Best Hypothesis to Explain the Data."
Learning problems are some unknown probability distribution D, over input-output pairs (x,y) ∈ XxY.
ML's Key Underlying Assumption
The synthesized and simplified representation of our training data.
QuAM or Model
A set of general instructions that builds a model of the target, using our learning data.
The Learning Algorithm