This paper presents a complicated problem decomposition-based approach called SWT

to the Boolean function minimization. SWT

comprises three major stages: First

a conceptual clustering system LEOBS partitions the given Boolean function into a number of subexpressions; then

the learning from examples based Boolean function minimization are system SWT minimizes each of the subexpressions; finally

the simplified subexpressions composed and the obtained function is further simplified globally. Some running results show that the proposed method is applicable to the large scale problems with more than 20 input variables.