Direct discriminative bag mapping for multi-instance learning

Jia Wu, Shirui Pan, Peng Zhang, Xingquan Zhu

Research output: Chapter in Book/Report/Conference proceedingConference PaperOtherpeer-review

Abstract

Multi-instance learning (MIL) is useful for tackling labeling ambiguity in learning tasks, by allowing a bag of instances to share one label. Recently, bag mapping methods, which transform a bag to a single instance in a new space via instance selection, have drawn significant attentions. To date, most existing works are developed based on the original space, i.e., utilizing all instances for bag mapping, and instance selection is indirectly tied to the MIL objective. As a result, it is hard to guarantee the distinguish capacity of the selected instances in the new bag mapping space for MIL. In this paper, we propose a direct discriminative mapping approach for multi-instance learning (MILDM), which identifies instances to directly distinguish bags in the new mapping space. Experiments and comparisons on real-world learning tasks demonstrate the algorithm performance.

Original languageEnglish
Title of host publicationProceedings of the Thirtieth AAAI Conference on Artificial Intelligence (AAAI'16 )
Subtitle of host publicationFebruary 12–17, 2016 Phoenix, Arizona, USA
EditorsDale Schuurmans, Michael Wellman
Place of PublicationPalo Alto CA USA
PublisherAssociation for the Advancement of Artificial Intelligence (AAAI)
Pages4274-4275
Number of pages2
ISBN (Electronic)9781577357605
Publication statusPublished - 2016
Externally publishedYes
EventAAAI Conference on Artificial Intelligence 2016 - Phoenix Convention Center, Phoenix, United States of America
Duration: 12 Feb 201617 Feb 2016
Conference number: 30th
http://www.aaai.org/Conferences/AAAI/aaai16.php

Conference

ConferenceAAAI Conference on Artificial Intelligence 2016
Abbreviated titleAAAI 16
CountryUnited States of America
CityPhoenix
Period12/02/1617/02/16
Internet address

Cite this

Wu, J., Pan, S., Zhang, P., & Zhu, X. (2016). Direct discriminative bag mapping for multi-instance learning. In D. Schuurmans, & M. Wellman (Eds.), Proceedings of the Thirtieth AAAI Conference on Artificial Intelligence (AAAI'16 ): February 12–17, 2016 Phoenix, Arizona, USA (pp. 4274-4275). [2818] Palo Alto CA USA: Association for the Advancement of Artificial Intelligence (AAAI).
Wu, Jia ; Pan, Shirui ; Zhang, Peng ; Zhu, Xingquan. / Direct discriminative bag mapping for multi-instance learning. Proceedings of the Thirtieth AAAI Conference on Artificial Intelligence (AAAI'16 ): February 12–17, 2016 Phoenix, Arizona, USA. editor / Dale Schuurmans ; Michael Wellman. Palo Alto CA USA : Association for the Advancement of Artificial Intelligence (AAAI), 2016. pp. 4274-4275
@inproceedings{1ec9d90a45bf44278aba038f51280297,
title = "Direct discriminative bag mapping for multi-instance learning",
abstract = "Multi-instance learning (MIL) is useful for tackling labeling ambiguity in learning tasks, by allowing a bag of instances to share one label. Recently, bag mapping methods, which transform a bag to a single instance in a new space via instance selection, have drawn significant attentions. To date, most existing works are developed based on the original space, i.e., utilizing all instances for bag mapping, and instance selection is indirectly tied to the MIL objective. As a result, it is hard to guarantee the distinguish capacity of the selected instances in the new bag mapping space for MIL. In this paper, we propose a direct discriminative mapping approach for multi-instance learning (MILDM), which identifies instances to directly distinguish bags in the new mapping space. Experiments and comparisons on real-world learning tasks demonstrate the algorithm performance.",
author = "Jia Wu and Shirui Pan and Peng Zhang and Xingquan Zhu",
year = "2016",
language = "English",
pages = "4274--4275",
editor = "Schuurmans, {Dale } and Wellman, {Michael }",
booktitle = "Proceedings of the Thirtieth AAAI Conference on Artificial Intelligence (AAAI'16 )",
publisher = "Association for the Advancement of Artificial Intelligence (AAAI)",
address = "United States of America",

}

Wu, J, Pan, S, Zhang, P & Zhu, X 2016, Direct discriminative bag mapping for multi-instance learning. in D Schuurmans & M Wellman (eds), Proceedings of the Thirtieth AAAI Conference on Artificial Intelligence (AAAI'16 ): February 12–17, 2016 Phoenix, Arizona, USA., 2818, Association for the Advancement of Artificial Intelligence (AAAI), Palo Alto CA USA, pp. 4274-4275, AAAI Conference on Artificial Intelligence 2016, Phoenix, United States of America, 12/02/16.

Direct discriminative bag mapping for multi-instance learning. / Wu, Jia; Pan, Shirui; Zhang, Peng; Zhu, Xingquan.

Proceedings of the Thirtieth AAAI Conference on Artificial Intelligence (AAAI'16 ): February 12–17, 2016 Phoenix, Arizona, USA. ed. / Dale Schuurmans; Michael Wellman. Palo Alto CA USA : Association for the Advancement of Artificial Intelligence (AAAI), 2016. p. 4274-4275 2818.

Research output: Chapter in Book/Report/Conference proceedingConference PaperOtherpeer-review

TY - GEN

T1 - Direct discriminative bag mapping for multi-instance learning

AU - Wu, Jia

AU - Pan, Shirui

AU - Zhang, Peng

AU - Zhu, Xingquan

PY - 2016

Y1 - 2016

N2 - Multi-instance learning (MIL) is useful for tackling labeling ambiguity in learning tasks, by allowing a bag of instances to share one label. Recently, bag mapping methods, which transform a bag to a single instance in a new space via instance selection, have drawn significant attentions. To date, most existing works are developed based on the original space, i.e., utilizing all instances for bag mapping, and instance selection is indirectly tied to the MIL objective. As a result, it is hard to guarantee the distinguish capacity of the selected instances in the new bag mapping space for MIL. In this paper, we propose a direct discriminative mapping approach for multi-instance learning (MILDM), which identifies instances to directly distinguish bags in the new mapping space. Experiments and comparisons on real-world learning tasks demonstrate the algorithm performance.

AB - Multi-instance learning (MIL) is useful for tackling labeling ambiguity in learning tasks, by allowing a bag of instances to share one label. Recently, bag mapping methods, which transform a bag to a single instance in a new space via instance selection, have drawn significant attentions. To date, most existing works are developed based on the original space, i.e., utilizing all instances for bag mapping, and instance selection is indirectly tied to the MIL objective. As a result, it is hard to guarantee the distinguish capacity of the selected instances in the new bag mapping space for MIL. In this paper, we propose a direct discriminative mapping approach for multi-instance learning (MILDM), which identifies instances to directly distinguish bags in the new mapping space. Experiments and comparisons on real-world learning tasks demonstrate the algorithm performance.

UR - http://www.scopus.com/inward/record.url?scp=85007268719&partnerID=8YFLogxK

M3 - Conference Paper

SP - 4274

EP - 4275

BT - Proceedings of the Thirtieth AAAI Conference on Artificial Intelligence (AAAI'16 )

A2 - Schuurmans, Dale

A2 - Wellman, Michael

PB - Association for the Advancement of Artificial Intelligence (AAAI)

CY - Palo Alto CA USA

ER -

Wu J, Pan S, Zhang P, Zhu X. Direct discriminative bag mapping for multi-instance learning. In Schuurmans D, Wellman M, editors, Proceedings of the Thirtieth AAAI Conference on Artificial Intelligence (AAAI'16 ): February 12–17, 2016 Phoenix, Arizona, USA. Palo Alto CA USA: Association for the Advancement of Artificial Intelligence (AAAI). 2016. p. 4274-4275. 2818