lec5

  Li near r egr essi on, acti ve learning We arriv ed at the lo gistic regression model when trying to explicitly model the uncertainty about the lab els in a linear c la ss ifier. The same genera l modeling approach p e rmits us to use line ar predictio ns in var ious other co ntexts as well. The simplest of them is regress ion where the go al is to pr e dict a con tin uous resp onse y t ∈ R to e ach example ve ctor. Here to o fo cusing on linear predictions won’t b e inherently limiting as linear predictions can b e easily extended (ne xt lecture).