Much like in other modelling disciplines does the distance metric used (a measure for dissimilarity) play an important role in the growing field of machine learning. Not surprisingly, in this field one also tries to learn the distance metric. In classification problems this has led to a dramatic performance boost. In this proposal we will
develop this learning methodology for ordinal classification problems (an important problem setting between classification and regression), with special attention for high-dimensional data, as they are often available nowadays.