Title :
Matching of attributed and nonattributed graphs by use of a Boltzmann machine algorithm
Author_Institution :
Corp. Res. & Dev., Siemens AG, Erlangen, West Germany
Abstract :
The paper deals with the task of solving the pure topological subgraph isomorphism problem and with the finding of a best relational matching subgraph to a given reference graph in an image graph with respect to both graphs´ attributes. In both cases, a combination of a simulated annealing strategy and of a Hungarian method algorithm is used. This combined strategy performs faster than the relaxation method and gives comparable results. It has two advantages. First, optimization over the states´ probability space is easier than generating the states themselves. Second, this principle renders very complex optimization possible by use of a simple and fast integer programming tool
Keywords :
computational complexity; graph theory; neural nets; optimisation; Boltzmann machine; Hungarian method algorithm; attributed graphs; integer programming tool; nonattributed graphs; optimization; pure topological subgraph isomorphism; relational matching subgraph; simulated annealing;
Conference_Titel :
Artificial Neural Networks, 1989., First IEE International Conference on (Conf. Publ. No. 313)
Conference_Location :
London