Skip to content
Free public beta until December 4, 2026: every course is open, no card needed. After that, the first chapter of each course stays free and full access is a one-time $39 ($19 for up to 100 founding members chosen for helpful feedback). Your progress is kept. How the beta works

stochastic gradient descent

Gradient descent whose gradient is estimated from a small random subset of the training data (a minibatch) instead of the whole dataset -- noisier per step, but far cheaper to compute.

Examples

  • On the five-car dataset, a minibatch of cars 1 and 4 gives gradient (-36, -6), close to but not identical to the full-dataset gradient (-38.4, -6).

See also

batch gradient descent
Read the full section →
Graph settings

Calculator