Communication Dans Un Congrès Année : 2018

Uplift Modeling from Separate Labels

Résumé

Uplift modeling is aimed at estimating the incremental impact of an action on an individual's behavior, which is useful in various application domains such as targeted marketing (advertisement campaigns) and personalized medicine (medical treatments). Conventional methods of uplift modeling require every instance to be jointly equipped with two types of labels: the taken action and its outcome. However, obtaining two labels for each instance at the same time is difficult or expensive in many real-world problems. In this paper, we propose a novel method of uplift modeling that is applicable to a more practical setting where only one type of labels is available for each instance. We show a mean squared error bound for the proposed estimator and demonstrate its effectiveness through experiments.

Mots clés

Fichier principal
Vignette du fichier
8198-uplift-modeling-from-separate-labels.pdf (2.29 Mo) Télécharger le fichier
Origine Fichiers produits par l'(les) auteur(s)
Licence
Loading...

Dates et versions

hal-02170693 , version 1 (02-07-2019)

Licence

Identifiants

  • HAL Id : hal-02170693 , version 1

Citer

Ikko Yamane, Florian Yger, Jamal Atif, Masashi Sugiyama. Uplift Modeling from Separate Labels. 32nd Conference on Neural Information Processing Systems (NeurIPS 2018), Dec 2018, Montréal, Canada. pp.9927--9937. ⟨hal-02170693⟩
57 Consultations
142 Téléchargements

Partager

  • More