1y

Loss Minimization

gradient descent towards consensus