SGD (Stochastic Gradient Descent)

A common optimiser that updates model parameters using small, randomly selected subsets (batches) of the training data, speeding up convergence.