Mathematics for ML

Gradient, explained

A gradient is a vector of partial derivatives showing how a function changes as its inputs change.
Back to Glossary

Part of the Semantic Notion technical glossary.

Short Definition

A gradient is a vector of partial derivatives showing how a function changes as its inputs change.

Intuition

The gradient points in the direction of steepest local increase.

Technical Definition

For a scalar-valued differentiable function, the gradient collects derivatives with respect to each input variable.

Example

Training uses the gradient of loss with respect to weights to decide parameter updates.

Common Misunderstandings

A gradient is local information, not a global map of the objective.
Small gradients can slow training even when better solutions exist.

Start here

Need the broader concept map?

Return to the glossary index to move from this definition into adjacent AI, ML, mathematics, and computer science terms.