We consider graph coloring problems where the cost of a coloring is the sum of the costs of the colors, and the cost of a color is a monotone concave function of the total weight of the class. This models resource allocation problems where the cost of a resource depends on the use of the resource. The specific case of interval graphs is of special interest as multi-criteria interval scheduling. We give an algorithm for all perfect graphs that yields a robust coloring: a particular solution that simultaneously approximates all concave functions. For graphs with uniform weights, we show how to modify the solution to approximate any monotone cost function. We complement these results with a number of hardness results and some exact algorithms on restricted classes of graphs.