Decomposition of multiple-valued relations

This paper presents a new decomposition problem: decomposition of multi-valued (MV) relations, and a method of its solution. Decomposition is non-disjoint and multi-level. A fundamental difference in decomposition of MV functions and MV relations is discussed: the column (cofactor) pair compatibility translates to the group compatibility for functions, but not for relations. This makes the decomposition of relations more difficult. The method is especially efficient for strongly unspecified data typical for Machine Learning (ML). It is implemented in program GUD-MV.(1)