Macsum Aggregation Learning and Missing Values

In recent work, a new kind of aggregation method has been proposed under the name of MacSum aggregation function that can be viewed as an interval valued aggregation function that is controlled by a precise vector of weights. This aggregation can be seen as a real valued extension of the possibility based aggregation. In this article, we show that a MacSum aggregation can be learned by using an input-output database where some input vectors have missing values.