Rawlsian Gradient Descent

Using Rawlsian Fairness for optimization