A discontinuous signum-function-activated (SFA) weights-and-structure-determination (WASD) neuronet model is presented and constructed based on the WASD algorithm. By this algorithm

the optimal weights and structure can be determined effectively. We apply the SFA-WASD neuronet model to XOR (i.e.

exclusive or)

and detail its performance in the XOR application with various types of disturbance noise considered. Numerical verification results substantiate the validity of the WASD algorithm in determining the optimal weights and structure

as well as the good anti-noise ability of the SFA-WASD neuronet in the XOR application. Moreover

for high-dimension XOR application

the performance comparison is made between the power-functionactivated (PFA) WASD neuronet and the SFA-WASD neuronet. The numerical results verify the superiority of the SFA-WASD neuronet in terms of solving nonlinear problems.