Late Fusion Incomplete Multi-view Clustering

Xinwang Liu, Xinzhong Zhu, Miaomiao Li, Lei Wang, Chang Tang, Jianping Yin, Dinggang Shen, Huaimin Wang, Wen Gao

Research output: Contribution to journalArticle

11 Citations (Scopus)

Abstract

Incomplete multi-view clustering optimally integrates a group of pre-specified incomplete views to improve clustering performance. Among various excellent solutions, multiple kernel <formula><tex>$k$</tex></formula>-means with incomplete kernels forms a benchmark, which redefines the incomplete multi-view clustering as a joint optimization problem where the imputation and clustering are alternately performed until convergence. However, the comparatively intensive computational and storage complexities preclude it from practical applications. To address these issues, we propose Late Fusion Incomplete Multi-view Clustering (LF-IMVC) which effectively and efficiently integrates the incomplete clustering matrices generated by incomplete views. Specifically, our algorithm jointly learns a consensus clustering matrix, imputes each incomplete base matrix, and optimizes the corresponding permutation matrices. We develop a three-step iterative algorithm to solve the resultant optimization problem with linear computational complexity and theoretically prove its convergence. Further, we conduct comprehensive experiments to study the proposed LF-IMVC in terms of clustering accuracy, running time, advantages of late fusion multi-view clustering, evolution of the learned consensus clustering matrix, parameter sensitivity and convergence. As indicated, our algorithm significantly and consistently outperforms some state-of-the-art algorithms with much less running time and memory.

Original languageEnglish
JournalIEEE Transactions on Pattern Analysis and Machine Intelligence
DOIs
Publication statusAccepted/In press - 2018 Jan 1
Externally publishedYes

Fingerprint

Cluster Analysis
Fusion
Fusion reactions
Clustering
Computational complexity
Consensus
Integrate
Data storage equipment
kernel
Optimization Problem
Benchmarking
Parameter Sensitivity
Permutation Matrix
Linear Complexity
Imputation
K-means
Multiple Solutions
Iterative Algorithm
Experiments
Computational Complexity

Keywords

  • incomplete kernel learning
  • multiple kernel clustering
  • multiple view learning

ASJC Scopus subject areas

  • Software
  • Computer Vision and Pattern Recognition
  • Computational Theory and Mathematics
  • Artificial Intelligence
  • Applied Mathematics

Cite this

Late Fusion Incomplete Multi-view Clustering. / Liu, Xinwang; Zhu, Xinzhong; Li, Miaomiao; Wang, Lei; Tang, Chang; Yin, Jianping; Shen, Dinggang; Wang, Huaimin; Gao, Wen.

In: IEEE Transactions on Pattern Analysis and Machine Intelligence, 01.01.2018.

Research output: Contribution to journalArticle

Liu, Xinwang ; Zhu, Xinzhong ; Li, Miaomiao ; Wang, Lei ; Tang, Chang ; Yin, Jianping ; Shen, Dinggang ; Wang, Huaimin ; Gao, Wen. / Late Fusion Incomplete Multi-view Clustering. In: IEEE Transactions on Pattern Analysis and Machine Intelligence. 2018.
@article{ad152b6123904f2db0588e369b85ff37,
title = "Late Fusion Incomplete Multi-view Clustering",
abstract = "Incomplete multi-view clustering optimally integrates a group of pre-specified incomplete views to improve clustering performance. Among various excellent solutions, multiple kernel $k$-means with incomplete kernels forms a benchmark, which redefines the incomplete multi-view clustering as a joint optimization problem where the imputation and clustering are alternately performed until convergence. However, the comparatively intensive computational and storage complexities preclude it from practical applications. To address these issues, we propose Late Fusion Incomplete Multi-view Clustering (LF-IMVC) which effectively and efficiently integrates the incomplete clustering matrices generated by incomplete views. Specifically, our algorithm jointly learns a consensus clustering matrix, imputes each incomplete base matrix, and optimizes the corresponding permutation matrices. We develop a three-step iterative algorithm to solve the resultant optimization problem with linear computational complexity and theoretically prove its convergence. Further, we conduct comprehensive experiments to study the proposed LF-IMVC in terms of clustering accuracy, running time, advantages of late fusion multi-view clustering, evolution of the learned consensus clustering matrix, parameter sensitivity and convergence. As indicated, our algorithm significantly and consistently outperforms some state-of-the-art algorithms with much less running time and memory.",
keywords = "incomplete kernel learning, multiple kernel clustering, multiple view learning",
author = "Xinwang Liu and Xinzhong Zhu and Miaomiao Li and Lei Wang and Chang Tang and Jianping Yin and Dinggang Shen and Huaimin Wang and Wen Gao",
year = "2018",
month = "1",
day = "1",
doi = "10.1109/TPAMI.2018.2879108",
language = "English",
journal = "IEEE Transactions on Pattern Analysis and Machine Intelligence",
issn = "0162-8828",
publisher = "IEEE Computer Society",

}

TY - JOUR

T1 - Late Fusion Incomplete Multi-view Clustering

AU - Liu, Xinwang

AU - Zhu, Xinzhong

AU - Li, Miaomiao

AU - Wang, Lei

AU - Tang, Chang

AU - Yin, Jianping

AU - Shen, Dinggang

AU - Wang, Huaimin

AU - Gao, Wen

PY - 2018/1/1

Y1 - 2018/1/1

N2 - Incomplete multi-view clustering optimally integrates a group of pre-specified incomplete views to improve clustering performance. Among various excellent solutions, multiple kernel $k$-means with incomplete kernels forms a benchmark, which redefines the incomplete multi-view clustering as a joint optimization problem where the imputation and clustering are alternately performed until convergence. However, the comparatively intensive computational and storage complexities preclude it from practical applications. To address these issues, we propose Late Fusion Incomplete Multi-view Clustering (LF-IMVC) which effectively and efficiently integrates the incomplete clustering matrices generated by incomplete views. Specifically, our algorithm jointly learns a consensus clustering matrix, imputes each incomplete base matrix, and optimizes the corresponding permutation matrices. We develop a three-step iterative algorithm to solve the resultant optimization problem with linear computational complexity and theoretically prove its convergence. Further, we conduct comprehensive experiments to study the proposed LF-IMVC in terms of clustering accuracy, running time, advantages of late fusion multi-view clustering, evolution of the learned consensus clustering matrix, parameter sensitivity and convergence. As indicated, our algorithm significantly and consistently outperforms some state-of-the-art algorithms with much less running time and memory.

AB - Incomplete multi-view clustering optimally integrates a group of pre-specified incomplete views to improve clustering performance. Among various excellent solutions, multiple kernel $k$-means with incomplete kernels forms a benchmark, which redefines the incomplete multi-view clustering as a joint optimization problem where the imputation and clustering are alternately performed until convergence. However, the comparatively intensive computational and storage complexities preclude it from practical applications. To address these issues, we propose Late Fusion Incomplete Multi-view Clustering (LF-IMVC) which effectively and efficiently integrates the incomplete clustering matrices generated by incomplete views. Specifically, our algorithm jointly learns a consensus clustering matrix, imputes each incomplete base matrix, and optimizes the corresponding permutation matrices. We develop a three-step iterative algorithm to solve the resultant optimization problem with linear computational complexity and theoretically prove its convergence. Further, we conduct comprehensive experiments to study the proposed LF-IMVC in terms of clustering accuracy, running time, advantages of late fusion multi-view clustering, evolution of the learned consensus clustering matrix, parameter sensitivity and convergence. As indicated, our algorithm significantly and consistently outperforms some state-of-the-art algorithms with much less running time and memory.

KW - incomplete kernel learning

KW - multiple kernel clustering

KW - multiple view learning

UR - http://www.scopus.com/inward/record.url?scp=85055896199&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85055896199&partnerID=8YFLogxK

U2 - 10.1109/TPAMI.2018.2879108

DO - 10.1109/TPAMI.2018.2879108

M3 - Article

AN - SCOPUS:85055896199

JO - IEEE Transactions on Pattern Analysis and Machine Intelligence

JF - IEEE Transactions on Pattern Analysis and Machine Intelligence

SN - 0162-8828

ER -