Multi-view Positive and Unlabeled Learning

Joey Tianyi Zhou, Sinno Jialin Pan, Qi Mao, Ivor W. Tsang
; Proceedings of the Asian Conference on Machine Learning, PMLR 25:555-570, 2012.

Abstract

Learning with Positive and Unlabeled instances (PU learning) arises widely in information retrieval applications. To address the unavailability issue of negative instances, most existing PU learning approaches require to either identify a reliable set of negative instances from the unlabeled data or estimate probability densities as an intermediate step. However, inaccurate negative-instance identification or poor density estimation may severely degrade overall performance of the final predictive model. To this end, we propose a novel PU learning method based on density ratio estimation without constructing any sets of negative instances or estimating any intermediate densities. To further boost PU learning performance, we extend our proposed learning method in a multi-view manner by utilizing multiple heterogeneous sources. Extensive experimental studies demonstrate the effectiveness of our proposed methods, especially when positive labeled data are limited.

Cite this Paper


BibTeX
@InProceedings{pmlr-v25-zhou12, title = {Multi-view Positive and Unlabeled Learning}, author = {Joey Tianyi Zhou and Sinno Jialin Pan and Qi Mao and Ivor W. Tsang}, booktitle = {Proceedings of the Asian Conference on Machine Learning}, pages = {555--570}, year = {2012}, editor = {Steven C. H. Hoi and Wray Buntine}, volume = {25}, series = {Proceedings of Machine Learning Research}, address = {Singapore Management University, Singapore}, month = {04--06 Nov}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v25/zhou12/zhou12.pdf}, url = {http://proceedings.mlr.press/v25/zhou12.html}, abstract = {Learning with Positive and Unlabeled instances (PU learning) arises widely in information retrieval applications. To address the unavailability issue of negative instances, most existing PU learning approaches require to either identify a reliable set of negative instances from the unlabeled data or estimate probability densities as an intermediate step. However, inaccurate negative-instance identification or poor density estimation may severely degrade overall performance of the final predictive model. To this end, we propose a novel PU learning method based on density ratio estimation without constructing any sets of negative instances or estimating any intermediate densities. To further boost PU learning performance, we extend our proposed learning method in a multi-view manner by utilizing multiple heterogeneous sources. Extensive experimental studies demonstrate the effectiveness of our proposed methods, especially when positive labeled data are limited.} }
Endnote
%0 Conference Paper %T Multi-view Positive and Unlabeled Learning %A Joey Tianyi Zhou %A Sinno Jialin Pan %A Qi Mao %A Ivor W. Tsang %B Proceedings of the Asian Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2012 %E Steven C. H. Hoi %E Wray Buntine %F pmlr-v25-zhou12 %I PMLR %J Proceedings of Machine Learning Research %P 555--570 %U http://proceedings.mlr.press %V 25 %W PMLR %X Learning with Positive and Unlabeled instances (PU learning) arises widely in information retrieval applications. To address the unavailability issue of negative instances, most existing PU learning approaches require to either identify a reliable set of negative instances from the unlabeled data or estimate probability densities as an intermediate step. However, inaccurate negative-instance identification or poor density estimation may severely degrade overall performance of the final predictive model. To this end, we propose a novel PU learning method based on density ratio estimation without constructing any sets of negative instances or estimating any intermediate densities. To further boost PU learning performance, we extend our proposed learning method in a multi-view manner by utilizing multiple heterogeneous sources. Extensive experimental studies demonstrate the effectiveness of our proposed methods, especially when positive labeled data are limited.
RIS
TY - CPAPER TI - Multi-view Positive and Unlabeled Learning AU - Joey Tianyi Zhou AU - Sinno Jialin Pan AU - Qi Mao AU - Ivor W. Tsang BT - Proceedings of the Asian Conference on Machine Learning PY - 2012/11/17 DA - 2012/11/17 ED - Steven C. H. Hoi ED - Wray Buntine ID - pmlr-v25-zhou12 PB - PMLR SP - 555 DP - PMLR EP - 570 L1 - http://proceedings.mlr.press/v25/zhou12/zhou12.pdf UR - http://proceedings.mlr.press/v25/zhou12.html AB - Learning with Positive and Unlabeled instances (PU learning) arises widely in information retrieval applications. To address the unavailability issue of negative instances, most existing PU learning approaches require to either identify a reliable set of negative instances from the unlabeled data or estimate probability densities as an intermediate step. However, inaccurate negative-instance identification or poor density estimation may severely degrade overall performance of the final predictive model. To this end, we propose a novel PU learning method based on density ratio estimation without constructing any sets of negative instances or estimating any intermediate densities. To further boost PU learning performance, we extend our proposed learning method in a multi-view manner by utilizing multiple heterogeneous sources. Extensive experimental studies demonstrate the effectiveness of our proposed methods, especially when positive labeled data are limited. ER -
APA
Zhou, J.T., Pan, S.J., Mao, Q. & Tsang, I.W.. (2012). Multi-view Positive and Unlabeled Learning. Proceedings of the Asian Conference on Machine Learning, in PMLR 25:555-570

Related Material