Abstract: Choosing how to encode a real-world problem as a machine learning task is an important design decision in machine learning. The task of the glacier calving front modeling has often been ...