Neural Networks · Level 2 of 5
Weight
A learned coefficient controlling an input's contribution to a computation.
Changing a weight changes how strongly a signal affects the output.
Example
A coefficient multiplies a feature before it enters a weighted sum.
Listen to the definition and example
Audio transcript
Weight. A learned coefficient controlling an input's contribution to a computation. Changing a weight changes how strongly a signal affects the output. For example: A coefficient multiplies a feature before it enters a weighted sum.
Explore this concept
A useful analogy
A volume control for one contribution entering a mixture.
Why it matters
This is part of the vocabulary used to read network diagrams and understand parameter updates.
Technical detail
For an affine unit z = Σ_i w_i x_i + b, w_i scales input x_i.
Common misconception
A large raw weight does not imply importance independently of feature scale and correlations.
Start with
Quick recall question
Try answering before looking back at the definition.