Multiple kernel-based multi-instance learning algorithm for image classification

Daxiang Li, Jing Wang, Xiaoqiang Zhao, Ying Liu, Dianwei Wang

Research output: Contribution to journalArticle

14 Citations (Scopus)

Abstract

In this paper, a novel multi-instance learning (MIL) algorithm based on multiple-kernels (MK) framework has been proposed for image classification. This newly developed algorithm defines each image as a bag, and the low-level visual features extracted from its segmented regions as instances. This algorithm is started from constructing a "word-space" from instances based on a collection of "visual-words" generated by affinity propagation (AP) clustering method. After calculating the distance between a "visual- word" and the bag (image), a nonlinear mapping mechanism is introduced for registering each bag as a coordinate point in the "word-space". In this case, the MIL problem is transformed into a standard supervised learning problem, which allows multiple-kernels support vector machine (MKSVM) classifiers to be trained for the image categorization. Compared with many popular MIL algorithms, the proposed method, named as MKSVM-MIL, shows its satisfactorily experimental results on the COREL dataset, which highlights the robustness and effectiveness for image classification applications.

Original languageEnglish
Pages (from-to)1112-1117
Number of pages6
JournalJournal of Visual Communication and Image Representation
Volume25
Issue number5
DOIs
Publication statusPublished - 2014

Fingerprint

Image classification
Learning algorithms
Support vector machines
Supervised learning
Classifiers

Cite this

Li, Daxiang ; Wang, Jing ; Zhao, Xiaoqiang ; Liu, Ying ; Wang, Dianwei. / Multiple kernel-based multi-instance learning algorithm for image classification. In: Journal of Visual Communication and Image Representation. 2014 ; Vol. 25, No. 5. pp. 1112-1117.
@article{42dc995bc49d4e058b6b7591489671a2,
title = "Multiple kernel-based multi-instance learning algorithm for image classification",
abstract = "In this paper, a novel multi-instance learning (MIL) algorithm based on multiple-kernels (MK) framework has been proposed for image classification. This newly developed algorithm defines each image as a bag, and the low-level visual features extracted from its segmented regions as instances. This algorithm is started from constructing a {"}word-space{"} from instances based on a collection of {"}visual-words{"} generated by affinity propagation (AP) clustering method. After calculating the distance between a {"}visual- word{"} and the bag (image), a nonlinear mapping mechanism is introduced for registering each bag as a coordinate point in the {"}word-space{"}. In this case, the MIL problem is transformed into a standard supervised learning problem, which allows multiple-kernels support vector machine (MKSVM) classifiers to be trained for the image categorization. Compared with many popular MIL algorithms, the proposed method, named as MKSVM-MIL, shows its satisfactorily experimental results on the COREL dataset, which highlights the robustness and effectiveness for image classification applications.",
keywords = "Affinity propagation (AP), Cluster analysis, Image classification, Image retrieval, Multi-instance learning (MIL), Multiple kernel learning (MKL), Support vector machines, Visual words",
author = "Daxiang Li and Jing Wang and Xiaoqiang Zhao and Ying Liu and Dianwei Wang",
year = "2014",
doi = "10.1016/j.jvcir.2014.03.011",
language = "English",
volume = "25",
pages = "1112--1117",
journal = "Journal of Visual Communication and Image Representation",
issn = "1047-3203",
publisher = "Academic Press Inc.",
number = "5",

}

Multiple kernel-based multi-instance learning algorithm for image classification. / Li, Daxiang; Wang, Jing; Zhao, Xiaoqiang; Liu, Ying; Wang, Dianwei.

In: Journal of Visual Communication and Image Representation, Vol. 25, No. 5, 2014, p. 1112-1117.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Multiple kernel-based multi-instance learning algorithm for image classification

AU - Li, Daxiang

AU - Wang, Jing

AU - Zhao, Xiaoqiang

AU - Liu, Ying

AU - Wang, Dianwei

PY - 2014

Y1 - 2014

N2 - In this paper, a novel multi-instance learning (MIL) algorithm based on multiple-kernels (MK) framework has been proposed for image classification. This newly developed algorithm defines each image as a bag, and the low-level visual features extracted from its segmented regions as instances. This algorithm is started from constructing a "word-space" from instances based on a collection of "visual-words" generated by affinity propagation (AP) clustering method. After calculating the distance between a "visual- word" and the bag (image), a nonlinear mapping mechanism is introduced for registering each bag as a coordinate point in the "word-space". In this case, the MIL problem is transformed into a standard supervised learning problem, which allows multiple-kernels support vector machine (MKSVM) classifiers to be trained for the image categorization. Compared with many popular MIL algorithms, the proposed method, named as MKSVM-MIL, shows its satisfactorily experimental results on the COREL dataset, which highlights the robustness and effectiveness for image classification applications.

AB - In this paper, a novel multi-instance learning (MIL) algorithm based on multiple-kernels (MK) framework has been proposed for image classification. This newly developed algorithm defines each image as a bag, and the low-level visual features extracted from its segmented regions as instances. This algorithm is started from constructing a "word-space" from instances based on a collection of "visual-words" generated by affinity propagation (AP) clustering method. After calculating the distance between a "visual- word" and the bag (image), a nonlinear mapping mechanism is introduced for registering each bag as a coordinate point in the "word-space". In this case, the MIL problem is transformed into a standard supervised learning problem, which allows multiple-kernels support vector machine (MKSVM) classifiers to be trained for the image categorization. Compared with many popular MIL algorithms, the proposed method, named as MKSVM-MIL, shows its satisfactorily experimental results on the COREL dataset, which highlights the robustness and effectiveness for image classification applications.

KW - Affinity propagation (AP)

KW - Cluster analysis

KW - Image classification

KW - Image retrieval

KW - Multi-instance learning (MIL)

KW - Multiple kernel learning (MKL)

KW - Support vector machines

KW - Visual words

UR - http://www.scopus.com/inward/record.url?scp=84899476100&partnerID=8YFLogxK

U2 - 10.1016/j.jvcir.2014.03.011

DO - 10.1016/j.jvcir.2014.03.011

M3 - Article

VL - 25

SP - 1112

EP - 1117

JO - Journal of Visual Communication and Image Representation

JF - Journal of Visual Communication and Image Representation

SN - 1047-3203

IS - 5

ER -