Improving Predictors via Combination Across Diverse Task Categories

                             Kwang In Kim 1

              Abstract                Predictor combination is related to multi-task learning
    Predictor combination is the problem of improv-      (MTL) and transfer learning (TL). However, since no access
    ing a task predictor using predictors of other       to the forms of the references are provided, existing MTL
    tasks when the forms of individual predictors are     and ...