Constrained clustering for data assignment problems with examples of module placement

The authors introduce a method that aims to avoid local optima and improve upon conventional solutions to data assignment problems, viewing them in the context of constrained clustering. The clustering algorithm is an extension of the deterministic annealing method which incorporates clustering constraints, initially generates soft data assignments, and then reduces the fuzziness of these assignments by lowering an effective temperature. In the low temperature limit, a legal data assignment solution is obtained, which seeks to minimize the desired objective function. The method was implemented on the two-dimensional module placement problem. For the minimum squared wire length objective, the algorithm outperformed several other methods on module placement examples from the literature