AI glossary

What is Gradient Descent?

The optimization method that trains neural networks by repeatedly nudging parameters opposite to the gradient of the loss.

The learning rate controls step size: too small is slow, too large diverges. Optimizers like Adam are refined versions.

Related terms