Some issues on the weights analysis in multi-objective decision making and multiple learning machines system

Some important issues on the determination of weights in multi-objective decision making (MODM) and multiple learning machines system (MLMS) are discussed: it is presented that determining weights is an important evaluation process; some drawbacks of several current methods of determining weights with the help of optimization are analyzed. Meanwhile, it is pointed out that it´s not overall and unreliable to regard the determination of weights as a single optimization problem based on limited data; Review and new research are carried out for general variable weights and its special case-the variable weights integration of modular neural networks; the inconsistency of weights is elementarily discussed and the reasons on this inconsistency are analyzed.