Curvature, and What Momentum Buys

Name: Curvature, and What Momentum Buys
Duration: 222 s
Description: An elongated quadratic bowl with gradient descent zig-zagging across the narrow axis while creeping along the flat one, the same run with a velocity vector accumulating down the valley, and the two error curves on a log axis separating at 165 steps against 42.

The condition number is the single number that predicts how slow plain descent will be, and momentum changes that prediction from κ to √κ - a claim precise enough to check to six decimal places.

IntermediateModule 225 min · 100 XP

An elongated quadratic bowl with gradient descent zig-zagging across the narrow axis while creeping along the flat one, the same run with a velocity vector accumulating down the valley, and the two error curves on a log axis separating at 165 steps against 42.

This is a premium lesson

← Back to Optimization for Learning