Cross-corpus speech emotion recognition based on transfer non-negative matrix factorization
详细信息 查看全文
- 作者：Peng Song ; a ; pengsong@ytu.edu.cn" class="auth_mail" title="E-mail the corresponding author ; pengsongseu@gmail.com" class="auth_mail" title="E-mail the corresponding author ; Wenming Zhengb ; wenming_zheng@seu.edu.cn" class="auth_mail" title="E-mail the corresponding author ; Shifeng Ouc ; ousfeng@ytu.edu.cn" class="auth_mail" title="E-mail the corresponding author ; Xinran Zhangb ; 230139080@seu.edu.cn" class="auth_mail" title="E-mail the corresponding author ; Yun Jind ; jiny@jsnu.edu.cn" class="auth_mail" title="E-mail the corresponding author ; Jinglei Liua ; jinglei_liu@sina.com" class="auth_mail" title="E-mail the corresponding author ; Yanwei Yua ; yuyanwei@ytu.edu.cn" class="auth_mail" title="E-mail the corresponding author
- 关键词：Speech emotion recognition ; Transfer learning ; Non-negative matrix factorization ; Dimension reduction
- 刊名：Speech Communication
- 出版年：2016
- 出版时间：October 2016
- 年：2016
- 卷：83
- 期：Complete
- 页码：34-41
- 全文大小：778 K
文摘
- •
In this paper, we investigate the NMF algorithms for cross-corpus speech emotion recognition.
- •
We propose a transfer NMF scheme for cross-corpus speech emotion recognition, in which the NMF and MMD algorithms are jointly optimized.
- •
Two novel transfer NMF approaches, called transfer graph regularized NMF (TGNMF) and transfer constrained NMF (TCNMF), are presented, respectively.
- •
The optimization algorithms for TGNMF and TCNMF approaches are presented in detail.