Fuzzy clustering of EEE components for space industry

Algorithms with greedy heuristic procedures for mixture probability distribution separation

“…A general formulation of the mixture distribution separation problem is as follows. Suppose that the distribution density on set X ⊂ R n has the form of a mixture of k distributions (we assume that the distributions are normal) [1,2,3]:…”

“…At the same time, the number of objects could be rather small for the efficient use of classical algorithms such as the EM algorithm [1]. Dividing a given lot of semiconductor devices shipped for usage in the space industry into a set of homogeneous production batches manufactured from homogeneous raw materials is a clustering problem (unsupervised learning) with rather high dimensionality of data vectors (up to thousands of dimensions) [1]. The number of clusters (homogeneous batches) is unknown, however, this number has known limitations.…”