This paper interprets an output of the single-layer perceptron as the synthetic evaluation of the similarity between the input pattern and two classes for two-class pattern classification problems. For this, an appropriate tool for multiple attribute decision making, TOPSIS, can be an appropriate tool. A novel single-layer perceptron based on the TOPSIS is proposed by measuring the grades of similarity between an input pattern and the typical patterns of the respective classes. To obtain the relative similarity in terms of certain distance measure, since the interaction among attributes cannot be ignored, the nonlinear Choquet integral with respect to the fuzzy measure is incorporated into the activation function. Moreover, a genetic algorithm-based method is designed to determine the connection weights, which are interpreted as the degrees of importance of the respective input signals. The experimental results further demonstrate that classification results obtained by the single-layer perceptron with the Choquet integral-based TOPSIS are comparable to those obtained by the other fuzzy or non-fuzzy classification methods.