wanghongjun
Research Associate
Supervisor of Master's Candidates
- Master Tutor
- Education Level:PhD graduate
- Degree:Doctor of engineering
- Business Address:犀浦3号教学楼31529
- Professional Title:Research Associate
- Alma Mater:四川大学
- Supervisor of Master's Candidates
- School/Department:计算机与人工智能学院
- Discipline:Electronic Information
Software Engineering
Computer Application Technology
Contact Information
- PostalAddress:
- Email:
- Paper Publications
Unsupervised Feature Learning Architecture with Multi-clustering Integration RBM
- Impact Factor:9.235
- DOI number:10.1109/TKDE.2020.3015959
- Affiliation of Author(s):西南交通大学
- Journal:IEEE Transactions on Knowledge and Data Engineering
- Place of Publication:UNITED STATES
- Key Words:Image reconstruction,Training,Feature extraction,Task analysis,Clustering algorithms,Machine learning,Encoding
- Abstract:In this paper, we present a novel unsupervised feature learning architecture, which consists of a multi-clustering integration module and a variant of RBM termed multi-clustering integration RBM (MIRBM). In the multi-clustering integration module, we apply three clusterers (K-means, affinity propagation and spectral clustering algorithms) to obtain three different clustering partitions (CPs) without any background knowledge or label. Then, an unanimous voting strategy is used to generate a local clustering partition (LCP). The novel MIRBM model is a core feature encoding part of the proposed unsupervised feature learning architecture. The novelty of it is that the LCP as an unsupervised guidance is integrated into one step contrastive divergence (CD1CD1) learning to guide the distribution of the hidden layer features. For the instance in the same LCP cluster, the hidden and reconstructed hidden layer features of the MIRBM model in the proposed architecture tend to constrict together in the training process. Meanwhile, each LCP center tends to disperse from each other as much as possible in the hidden and reconstructed hidden layer during training. The experiments demonstrate that the proposed unsupervised feature learning architecture has more powerful feature representation and generalization capability than the state-of-the-art models for clustering tasks in the Microsoft Research Asia Multimedia (MSRA-MM)2.0 dataset.
- Co-author:Jing Liu, Zhiguo Gong,Tianrui Li
- First Author:Jielei Chu
- Indexed by:Academic papers
- Correspondence Author:HongjunWang
- Document Code:20203809195020
- Discipline:Engineering
- First-Level Discipline:Computer Science and Technology
- Volume:Volume: 34
- Page Number:3002 - 3015
- ISSN No.:1041-4347
- Translation or Not:no
- Date of Publication:2022-06-01