Machine Learning TheoryCan the following loss function be used with gradient descent to learn a perceptron?

$E(w)=\frac{1}{N}{\displaystyle \underset{i=1}{\overset{n}{}}}$max$(0,-y_i{w}^T{x}_i)$