Keras: calculating loss as * median * via datapoints instead of average

The Keras loss page says that if we have a custom loss function, then "the actual optimized target is the average of the output array across all data points." Is there any way to optimize the median of the output array across all data points (instead of the average)?