Supervised manifold learning for image and video classification

Yang Liu; Yan Liu; Keith C.C. Chan

doi:10.1145/1873951.1874097

Supervised manifold learning for image and video classification

Yang Liu^*
, Yan Liu^*
, Keith C.C. Chan

^*Corresponding author for this work

Department of Computer Science

Research output: Chapter in book/report/conference proceeding › Conference proceeding › peer-review

5 Citations (Scopus)

Abstract

This paper presents a supervised manifold learning model for dimensionality reduction in image and video classification tasks. Unlike most manifold learning models that emphasize the distance preserving, we propose a novel algorithm called maximum distance embedding (MDE), which aims to maximize the distances between some particular pairs of data points, with the intention of flattening the local nonlinearity and keeping the discriminant information simultaneously in the embedded feature space. Moreover, MDE measures the dissimilarity between data points using L1-norm distance, which is more robust to outliers than widely used Frobenius norm distance. To adapt the nature tensor structure of image and video data, we further propose the multilinear MDE (M2DE). Experiments on various datasets demonstrate that both MDE and M2DE achieve impressive embedding results of image and video data for classification tasks.

Original language	English
Title of host publication	MM 2010 - Proceedings of the ACM Multimedia 2010 International Conference
Publisher	Association for Computing Machinery (ACM)
Pages	859-862
Number of pages	4
ISBN (Print)	9781605589336
DOIs	https://doi.org/10.1145/1873951.1874097
Publication status	Published - 25 Oct 2010
Event	18th ACM International Conference on Multimedia ACM Multimedia, MM 2010 - Firenze, Italy Duration: 25 Oct 2010 → 29 Oct 2010 https://dl.acm.org/doi/proceedings/10.1145/1873951 (Conference proceeding)

Publication series

Name	Proceedings of the ACM Multimedia International Conference

Conference

Conference	18th ACM International Conference on Multimedia ACM Multimedia, MM 2010
Country/Territory	Italy
City	Firenze
Period	25/10/10 → 29/10/10
Internet address	https://dl.acm.org/doi/proceedings/10.1145/1873951 (Conference proceeding)

UN SDGs

This output contributes to the following UN Sustainable Development Goals (SDGs)

SDG 9 Industry, Innovation, and Infrastructure

User-Defined Keywords

image and video classification
l1-norm optimization
manifold learning
maximum distance embedding
multilinear maximum distance embedding
supervised learning

Access to Document

10.1145/1873951.1874097

Cite this

@inproceedings{0dec19751bcc4700a9fbc3bd16660964,

title = "Supervised manifold learning for image and video classification",

abstract = "This paper presents a supervised manifold learning model for dimensionality reduction in image and video classification tasks. Unlike most manifold learning models that emphasize the distance preserving, we propose a novel algorithm called maximum distance embedding (MDE), which aims to maximize the distances between some particular pairs of data points, with the intention of flattening the local nonlinearity and keeping the discriminant information simultaneously in the embedded feature space. Moreover, MDE measures the dissimilarity between data points using L1-norm distance, which is more robust to outliers than widely used Frobenius norm distance. To adapt the nature tensor structure of image and video data, we further propose the multilinear MDE (M2DE). Experiments on various datasets demonstrate that both MDE and M2DE achieve impressive embedding results of image and video data for classification tasks.",

keywords = "image and video classification, l1-norm optimization, manifold learning, maximum distance embedding, multilinear maximum distance embedding, supervised learning",

author = "Yang Liu and Yan Liu and Chan, \{Keith C.C.\}",

note = "Publisher Copyright: {\textcopyright} 2010 ACM.; 18th ACM International Conference on Multimedia ACM Multimedia, MM 2010 ; Conference date: 25-10-2010 Through 29-10-2010",

year = "2010",

month = oct,

day = "25",

doi = "10.1145/1873951.1874097",

language = "English",

isbn = "9781605589336",

series = "Proceedings of the ACM Multimedia International Conference",

publisher = "Association for Computing Machinery (ACM)",

pages = "859--862",

booktitle = "MM 2010 - Proceedings of the ACM Multimedia 2010 International Conference",

address = "United States",

url = "https://dl.acm.org/doi/proceedings/10.1145/1873951",

}

Liu, Y, Liu, Y & Chan, KCC 2010, Supervised manifold learning for image and video classification. in MM 2010 - Proceedings of the ACM Multimedia 2010 International Conference. Proceedings of the ACM Multimedia International Conference, Association for Computing Machinery (ACM), pp. 859-862, 18th ACM International Conference on Multimedia ACM Multimedia, MM 2010, Firenze, Italy, 25/10/10. https://doi.org/10.1145/1873951.1874097

Supervised manifold learning for image and video classification. / Liu, Yang; Liu, Yan; Chan, Keith C.C.
MM 2010 - Proceedings of the ACM Multimedia 2010 International Conference. Association for Computing Machinery (ACM), 2010. p. 859-862 (Proceedings of the ACM Multimedia International Conference).

Research output: Chapter in book/report/conference proceeding › Conference proceeding › peer-review

TY - GEN

T1 - Supervised manifold learning for image and video classification

AU - Liu, Yang

AU - Liu, Yan

AU - Chan, Keith C.C.

PY - 2010/10/25

Y1 - 2010/10/25

N2 - This paper presents a supervised manifold learning model for dimensionality reduction in image and video classification tasks. Unlike most manifold learning models that emphasize the distance preserving, we propose a novel algorithm called maximum distance embedding (MDE), which aims to maximize the distances between some particular pairs of data points, with the intention of flattening the local nonlinearity and keeping the discriminant information simultaneously in the embedded feature space. Moreover, MDE measures the dissimilarity between data points using L1-norm distance, which is more robust to outliers than widely used Frobenius norm distance. To adapt the nature tensor structure of image and video data, we further propose the multilinear MDE (M2DE). Experiments on various datasets demonstrate that both MDE and M2DE achieve impressive embedding results of image and video data for classification tasks.

AB - This paper presents a supervised manifold learning model for dimensionality reduction in image and video classification tasks. Unlike most manifold learning models that emphasize the distance preserving, we propose a novel algorithm called maximum distance embedding (MDE), which aims to maximize the distances between some particular pairs of data points, with the intention of flattening the local nonlinearity and keeping the discriminant information simultaneously in the embedded feature space. Moreover, MDE measures the dissimilarity between data points using L1-norm distance, which is more robust to outliers than widely used Frobenius norm distance. To adapt the nature tensor structure of image and video data, we further propose the multilinear MDE (M2DE). Experiments on various datasets demonstrate that both MDE and M2DE achieve impressive embedding results of image and video data for classification tasks.

KW - image and video classification

KW - l1-norm optimization

KW - manifold learning

KW - maximum distance embedding

KW - multilinear maximum distance embedding

KW - supervised learning

UR - https://www.scopus.com/pages/publications/78650976368

U2 - 10.1145/1873951.1874097

DO - 10.1145/1873951.1874097

M3 - Conference proceeding

AN - SCOPUS:78650976368

SN - 9781605589336

T3 - Proceedings of the ACM Multimedia International Conference

SP - 859

EP - 862

BT - MM 2010 - Proceedings of the ACM Multimedia 2010 International Conference

PB - Association for Computing Machinery (ACM)

T2 - 18th ACM International Conference on Multimedia ACM Multimedia, MM 2010

Y2 - 25 October 2010 through 29 October 2010

ER -

Supervised manifold learning for image and video classification

Abstract

Publication series

Conference

UN SDGs

User-Defined Keywords

Access to Document

Other files and links

Fingerprint

Cite this