Energy-minimizing neural network with symbol-valued units and its application to the analysis of basal cell carcinoma

The energy-minimizing neural network with symmetrical Q-state neurons is described. The common operation of the network is formulated on the basis of probabilistic interpretation of network state energy. System performance is investigated on simulations of model data processing. The ability of the network to make effective extraction of signals from noisy data is used to process information about patients suffering basal cell carcinoma. The possibility of predicting relapses for these patients is also demonstrated