Induced aggregation operators

The induced ordered weighted averaging (IOWA) operator is introduced and it is shown its possibilities in modeling nearest-neighbor rules and also used to establish a new class of information fusion models called "best yesterday models".

Abstract

We introduce the induced ordered weighted averaging (IOWA) operator. In these operators the argument ordering process is guided by a variable called the order inducing value. A procedure for learning the weights from data is described. We suggest a number of applications of these induced OWA aggregation operators. First we show its possibilities in modeling nearest-neighbor rules. Next it is applied to the aggregation of complex objects such as matrices. It is also used to establish a new class of information fusion models called "best yesterday models". Finally, we extend the idea of order induced aggregation to the Choquet aggregation resulting in what we call the induced Choquet ordered averaging (I-COA) operator.