English: A plot showing gradient descent using arrows; the arrow size gets progressively smaller as the algorithm begins to converge. In this case, it is converging on two parameters for linear regression, attempting to minimize the cost function, ${\displaystyle J({\theta }_0,{\theta }_1)}$. The colour of the arrows has no significance; they simply alternate for visibility.