
软件学报 ISSN 1000-9825, CODEN RUXUEW E-mail: jos@iscas.ac.cn
Journal of Software, 2022,33(4):1373−1389 [doi: 10.13328/j.cnki.jos.006474] http://www.jos.org.cn
©中国科学院软件研究所版权所有. Tel: +86-10-62562563
CMvSC: 知识迁移下的深度一致性多视图谱聚类网络
∗
张熠玲
,
杨
燕
,
周
威
,
欧阳小草
,
胡
节
(西南交通大学 计算机与人工智能学院, 四川 成都 611756)
通信作者: 杨燕, E-mail: yyang@swjtu.edu.cn
摘 要: 谱聚类是聚类分析中极具代表性的方法之一, 由于其对数据结构没有太多假设要求, 受到了研究者们的
广泛关注. 但传统的谱聚类算法通常受到谱嵌入的可扩展性和泛化性的限制, 即: 无法应对大规模设置和复杂数
据分布. 为克服以上缺陷, 旨在引入深度学习框架提升谱聚类的泛化能力与可扩展能力, 同时, 结合多视图学习
挖掘数据样本的多样性特征, 从而提出一种知识迁移下的深度一致性多视图谱聚类网络(CMvSC). 首先, 考虑到
单个视图的局部不变性, CMvSC 采用局部学习层独立学习每个视图的特有嵌入; 其次, 由于多视图具有全局一致
性, CMvSC 引入全局学习层进行参数共享与特征迁移, 学习多视图间的共享嵌入; 同时, 考虑到邻接矩阵对谱聚
类性能的重要影响, CMvSC 通过训练孪生网络和设计对比损失来学习成对数据间的近邻关系, 以替代传统谱聚类
算法中的距离度量; 最后, 4 个数据集上的实验结果证明了 CMvSC 对多视图谱聚类任务的有效性.
关键词: 谱嵌入; 近邻学习; 知识迁移; 多视图聚类; 深度聚类
中图法分类号: TP181
中文引用格式: 张熠玲, 杨燕, 周威, 欧阳小草, 胡节. CMvSC: 知识迁移下的深度一致性多视图谱聚类网络. 软件学报,
2022, 33(4): 1373–1389. http://www.jos.org.cn/1000-9825/6474 .htm
英文引用格式: Zhang YL, Yang Y, Zhou W, Ouyang XC, Hu J. CMvSC: Knowledge Transferring Based Deep Consensus Network
for Multi-view Spectral Clustering. Ruan Jian Xue Bao/Journal of Software, 2022, 33(4): 1373−1389 (in Chinese). http://www.jos.
org.cn/ 1000-9825/6474.htm
CMvSC: Knowledge Transferring Based Deep Consensus Network for Multi-view Spectral
Clustering
ZHANG Yi-Ling, YANG Yan, ZHOU Wei, OUYANG Xiao-Cao, HU Jie
(School of Computing and Artifici al Intellig ence, South west Jiaotong University, Chengdu 611756, Chin a)
Abstra ct : Spectral clustering, which is one of the most representative methods in clustering analysis, receives much attention from
scholars, because it does not constrain th e data structure of the origin al samples. However, traditional spectral clustering algorithm usually
contains two major limitations, i.e., it is unable to cope with th e large-scale settings and complex dat a distribution. To overcome the above
shortcomings, this study introduces a deep learning framework to improve the generalization and scalability of spectral clustering, and
combines the multi-view learning to mine diverse features among data samples, finally proposes a knowledge transferring based deep
consensus network for multi-v iew spectral clustering (CMvSC). First , considering the local invariance of sing le view, CMvSC adopts the
local learning layer to learn the specific embedding of each view individually. Then, because of the global consistency among multiple
views, CMvSC introduces the glob al learning layer to achieve p arameter sh aring and f eatur e transferring, and l earns the sh ared embedding
in different views. Meanwhile, taking the effect of affinity matrix for spectral clustering into consideration, CMvSC learns the affinity
correlation between the paired samples by training the Siamese network and designing the contrastive loss, which replaces the distance
metric in traditional spectral clustering. Finally, the experimental results on four datasets demonstrate the effectiveness of the proposed
∗ 基金项目: 国家自然科学基金(61976 247)
本文由“面向开放场景的鲁棒机器学习”专刊特约编辑陈恩红教授、李宇峰副教授、邹权教授推荐.
收稿时间: 2021-05-29; 修改时间: 2021-07-16; 采用时间: 2021-08-27; jos 在线出版时间: 2 021-10-26
评论