deep-learning

The batch dimension and broadcasting, as engineering convention

taught

Stacking many examples along a leading axis, and the library rule that lets a smaller shape combine with a larger one -- neither is a fact of linear algebra itself.

Taught in

Shapes and the batch dimension
Deep Learning / Neural Networks
Broadcasting: a library convention, not a math rule
Deep Learning / Neural Networks

Requires

matrices-notation-shape-and-special-forms required
matrix-vector-multiplication-two-ways-the-row-picture helpful
Seeing one row's own computation first is what makes "do it once per row" click.

Required by

NumPy's broadcasting rule, mechanically required

Local knowledge map

Click to expand

Solid arrows are required prerequisites; dashed are recommended or helpful. Lower rows are more foundational. Click any box to jump to that concept.