To deal with the problem of incomplete comparisons in decision maker preference assessment, a SONIA (Self-Organized Network inspired by Immune Algorithm)-based Decision Neural Network (DNN) is proposed. B cell mutation of SONIA creates hidden units having diverse data characteristics that improve generalization capability. This mutation deals with a limited number of training data resulting from incomplete pair-wise comparisons by decision maker. Experiments on linear, Euclidean, and L_p-metric function as underlying decision maker preference show that SONIA-based DNN outperforms conventional DNN for decision maker preference assessment with incomplete comparisons.