Machine Learning

Momentum

An optimization technique that accelerates gradient descent by accumulating a velocity vector in the direction of persistent gradients, helping overcome local minima and noisy gradients.

Why It Matters

Momentum helps models train faster and more stably by smoothing out the optimization path and building speed in consistent gradient directions.

Example

Without momentum, gradient descent oscillates in narrow valleys. With momentum, it builds speed in the downhill direction and dampens oscillations.

Think of it like...

Like a bowling ball rolling downhill — it builds up speed in a consistent direction and is not easily knocked off course by small bumps.

Related Terms