liveBook · Manning

Hyperplane

As explained in Grokking Machine Learning, Second Edition

A flat geometric boundary of one dimension less than its surrounding space that separates regions in that space.

A hyperplane is a flat boundary one dimension below the space around it. A line divides a plane; a plane divides three-dimensional space; in higher dimensions, a hyperplane divides feature space into two regions. section 5.9

Sentences as points

Represent each word with a coordinate. A sentence containing “aack” five times, “beep” eight times, and “crack” three times becomes the point (5, 8, 3). A perceptron’s plane can then split such points into two predicted classes, such as happy and sad. section 5.9

A perceptron uses weights and a bias to place this boundary. Points on one side receive one prediction, while points on the other side receive the other. The perceptron algorithm gradually moves the boundary when it misclassifies a point. section 5.27 section 5.9

Figure 5.10 A dataset of sentences with three words is plotted in three dimensions. The classifier is represented by a plane that splits the space into two regions.

Use the three-dimensional figure to see how a classifier’s plane separates a dataset into two regions. section 5.9

How Grokking Machine Learning, Second Edition covers it

How deep it goes
Defines the boundary through lines, planes, and higher-dimensional spaces, then ties it to weights and bias without deriving its equation.
Examples
Uses word-count coordinates, sentiment scores, and a three-dimensional sentence dataset split by a plane.
What it assumes
Builds dimensional intuition from points, lines, and planes before extending the idea to features and labels in n-dimensional space.
What you can do after
Understand how a perceptron separates feature points and why its boundary becomes difficult to visualize in many dimensions.
How it connects
Connects hyperplanes to perceptrons, weights, bias, features, labels, and classification.

Also explained in

Compare how 8 books explain hyperplane →

Related concepts

This explainer was generated by Manning's knowledge graph from the text of Grokking Machine Learning, Second Edition. The section links point to the parts of the book it draws on.