liveBook · Manning

Data augmentation

As explained in Timeless Algorithms, The Foundational Ideas

Data augmentation creates additional training observations by transforming existing observations while preserving their correct outcomes.

Data augmentation creates additional training observations by transforming existing ones while preserving their correct outcomes. The added variation encourages a network to learn predictive patterns that persist across changing conditions instead of memorizing particular examples. section 11.6.2

A pedestrian in changing images

Cropping, translating, reflecting, or changing the brightness of an image can produce additional examples that still depict the same pedestrian. section 11.6.2

Data augmentation constrains how a network uses its capacity, supporting the broader objective of generalization. section 11.8

How Timeless Algorithms, The Foundational Ideas covers it

How deep it goes
Defines label-preserving transformations and explains their role in learning predictive patterns across changing conditions.
Examples
Uses pedestrian images transformed by cropping, translation, reflection, or brightness changes.
What you can do after
Understand how label-preserving variation can discourage memorization of particular training examples.
How it connects
Links data augmentation to regularization, generalization, weight decay, dropout, early stopping, support vector machines, and random forests.

Also explained in

Compare how 12 books explain data augmentation →

Related concepts

This explainer was generated by Manning's knowledge graph from the text of Timeless Algorithms, The Foundational Ideas. The section links point to the parts of the book it draws on.