A quantum-inspired self-organizing map (QISOM)

A variation of the Kohonen map, the quantum-inspired self-organizing map (QISOM) is a topological map of "local" quantum states. The two-dimensional map is initialized to random quantum states with its training data and the training process selects each winner using a decoherence operator that reduces the quantum states to classic non-probabilistic states. The QISOM is not used to cluster new data; it is used as a current state of the population that is updated in real time. The QISOM is useful when observations from a very large population are presented in small batches. The result is that the prevalence of very rare events may be more accurate than with traditional statistical sampling in batches. However, this is accomplished at the cost of the absolute prevalence accuracy of more common events, although rank ordering of prevalence is preserved. The assumption for employing a QISOM is that the primary concern is to capture the prevalence of very rare critical and costly events. The QISOM is demonstrated with data on selected health and social problems of children. Results are compared with bootstrapping