Supervised Disentangled Graph Learning for OCM

Weili Guan; Xuemeng Song; Xiaojun Chang; Liqiang Nie

doi:10.1007/978-3-031-18817-6_5

Supervised Disentangled Graph Learning for OCM

Weili Guan
, Xuemeng Song
, Xiaojun Chang
, Liqiang Nie

Department of Data Science & AI

Research output: Chapter in Book/Report/Conference proceeding › Chapter (Book) › Other › peer-review

Abstract

In Chap. 4, we studied the fine-grained outfit compatibility modeling, where the hidden factors affecting the outfit compatibility are jointly considered. One key limitation is that it only investigates the visual content of fashion items while overlooking the items’ semantic attributes. The item attribute labels usually contain rich information that characterizes the key item parts, which can be adopted to supervise the attribute-level representation learning, and hence promote the model’s performance as well as interpretability. Thus, in this chapter, we aim to fulfill the fine-grained outfit compatibility modeling by incorporating the semantic attributes of fashion items.

Original language	English
Title of host publication	Graph Learning for Fashion Compatibility Modeling
Editors	Weili Guan, Xuemeng Song, Xiaojun Chang, Liqiang Nie
Place of Publication	Cham Switzerland
Publisher	Springer
Chapter	5
Pages	67-87
Number of pages	21
Edition	2nd
ISBN (Electronic)	9783031188176
ISBN (Print)	9783031188169
DOIs	https://doi.org/10.1007/978-3-031-18817-6_5
Publication status	Published - 2022

Publication series

Name	Synthesis Lectures on Information Concepts, Retrieval, and Services
Publisher	Springer Nature
ISSN (Print)	1947-945X
ISSN (Electronic)	1947-9468

Access to Document

10.1007/978-3-031-18817-6_5

Cite this

@inbook{baee44fb8bbd4862879c8987cb22dc5e,

title = "Supervised Disentangled Graph Learning for OCM",

abstract = "In Chap. 4, we studied the fine-grained outfit compatibility modeling, where the hidden factors affecting the outfit compatibility are jointly considered. One key limitation is that it only investigates the visual content of fashion items while overlooking the items{\textquoteright} semantic attributes. The item attribute labels usually contain rich information that characterizes the key item parts, which can be adopted to supervise the attribute-level representation learning, and hence promote the model{\textquoteright}s performance as well as interpretability. Thus, in this chapter, we aim to fulfill the fine-grained outfit compatibility modeling by incorporating the semantic attributes of fashion items.",

author = "Weili Guan and Xuemeng Song and Xiaojun Chang and Liqiang Nie",

note = "Publisher Copyright: {\textcopyright} 2022, The Author(s), under exclusive license to Springer Nature Switzerland AG.",

year = "2022",

doi = "10.1007/978-3-031-18817-6\_5",

language = "English",

isbn = "9783031188169",

series = "Synthesis Lectures on Information Concepts, Retrieval, and Services",

publisher = "Springer",

pages = "67--87",

editor = "Weili Guan and Xuemeng Song and Xiaojun Chang and Liqiang Nie",

booktitle = "Graph Learning for Fashion Compatibility Modeling",

address = "Switzerland",

edition = "2nd",

}

Supervised Disentangled Graph Learning for OCM. / Guan, Weili; Song, Xuemeng; Chang, Xiaojun et al.
Graph Learning for Fashion Compatibility Modeling. ed. / Weili Guan; Xuemeng Song; Xiaojun Chang; Liqiang Nie. 2nd. ed. Cham Switzerland: Springer, 2022. p. 67-87 (Synthesis Lectures on Information Concepts, Retrieval, and Services).

Research output: Chapter in Book/Report/Conference proceeding › Chapter (Book) › Other › peer-review

TY - CHAP

T1 - Supervised Disentangled Graph Learning for OCM

AU - Guan, Weili

AU - Song, Xuemeng

AU - Chang, Xiaojun

AU - Nie, Liqiang

PY - 2022

Y1 - 2022

N2 - In Chap. 4, we studied the fine-grained outfit compatibility modeling, where the hidden factors affecting the outfit compatibility are jointly considered. One key limitation is that it only investigates the visual content of fashion items while overlooking the items’ semantic attributes. The item attribute labels usually contain rich information that characterizes the key item parts, which can be adopted to supervise the attribute-level representation learning, and hence promote the model’s performance as well as interpretability. Thus, in this chapter, we aim to fulfill the fine-grained outfit compatibility modeling by incorporating the semantic attributes of fashion items.

AB - In Chap. 4, we studied the fine-grained outfit compatibility modeling, where the hidden factors affecting the outfit compatibility are jointly considered. One key limitation is that it only investigates the visual content of fashion items while overlooking the items’ semantic attributes. The item attribute labels usually contain rich information that characterizes the key item parts, which can be adopted to supervise the attribute-level representation learning, and hence promote the model’s performance as well as interpretability. Thus, in this chapter, we aim to fulfill the fine-grained outfit compatibility modeling by incorporating the semantic attributes of fashion items.

UR - https://www.scopus.com/pages/publications/85141831939

U2 - 10.1007/978-3-031-18817-6_5

DO - 10.1007/978-3-031-18817-6_5

M3 - Chapter (Book)

AN - SCOPUS:85141831939

SN - 9783031188169

T3 - Synthesis Lectures on Information Concepts, Retrieval, and Services

SP - 67

EP - 87

BT - Graph Learning for Fashion Compatibility Modeling

A2 - Guan, Weili

A2 - Song, Xuemeng

A2 - Chang, Xiaojun

A2 - Nie, Liqiang

PB - Springer

CY - Cham Switzerland

ER -

Supervised Disentangled Graph Learning for OCM

Abstract

Publication series

Access to Document

Other files and links

Research output

Graph Learning for Fashion Compatibility Modeling

Cite this