AI Glossary

Weight initialization

What does “Weight initialization” mean?

Weight initialization: Choosing a network's starting parameters before training.

Weight initialization visual guide: Choosing a network's starting parameters before training.
Visual guide Open full size (new tab)

What is an example of “Weight initialization”?

A network starts with small random weights so its units can learn different features.

Keep learning

See also

Deep learningTermTrainingTermOverfittingTermLayer normalizationTermRMS normalizationTerm

Same topic

Receptive fieldTermResidual connectionTermWeight decayTermWeak supervisionTermValue functionTerm