liveBook · Manning

Width

As explained in Timeless Algorithms, The Foundational Ideas

Network width is the number of computational units in a neural network layer.

A layer’s width is its number of nodes: width of layer l = number of nodes in layer l. Wider layers can represent more patterns at the same processing stage; depth instead composes representations across layers. section 11.2.1

In pedestrian recognition, different nodes may respond to visual patterns such as edges, textures, shapes, and contextual cues. Evidence is distributed across node responses. Width accommodates pattern variety, but capacity alone does not ensure generalization to unseen observations. section 11.5.1

How Timeless Algorithms, The Foundational Ideas covers it

How deep it goes
Defines width as nodes per layer and explains its role in representing varied visual patterns.
Examples
Uses pedestrian recognition to illustrate nodes responding to edges, textures, shapes, body parts, and context.
What you can do after
Understand how widening a layer increases pattern capacity without guaranteeing generalization.
How it connects
Connects width to depth, learned representations, neural-network capacity, and generalization.

Also explained in

Compare how 4 books explain width →

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.