Clustering Algorithm with a Greedy Agglomerative Heuristic and Special Distance Measures

Automatic grouping (clustering) involves dividing a set of objects into subsets (groups) so that the from one subset are more similar to each other than according some criterion. Kohonen neural networks class artificial networks, main element which is layer adaptive linear adders, operating on principle “winner takes all”. One advantages their ability online clustering. Greedy agglomerative procedures in clustering consistently improve result neighborhood known solution, choosing as next solution option provides least increase objective function. Algorithms using greedy heuristics demonstrate precise and stable results for k-means model. In our study, we propose heuristic algorithm based network with distance measure variations cluster industrial products. Computational experiments comparative efficiency accuracy problem products homogeneous production batches.