Convex hulls as boundary descriptors often become infeasible beyond ten dimensions; common alternatives include ellipsoids, hyperrectangles, or decision boundaries from support vector machines.