Лексикографический подход к оптимизации множества кусочно-линейных ограничений

The problem of piecewise-linear constraints reconstruction arising in optimization problems with incomplete data is considered in the paper. To reconstruct unknown constraints a perceptron is used. The lexicographical method is proposed to optimize obtained solution. A square matrix is built on the base of initial data and perceptron outputs, which is optimized due to the extraction of comparable row vector pairs and crossing out matrix rows and columns, corresponding to the lower vectors. The corresponding perceptron outputs are deleted too.