Z
Zhi-Hua Zhou
Researcher at Nanjing University
Publications - 633
Citations - 64307
Zhi-Hua Zhou is an academic researcher from Nanjing University. The author has contributed to research in topics: Semi-supervised learning & Artificial neural network. The author has an hindex of 102, co-authored 626 publications receiving 52850 citations. Previous affiliations of Zhi-Hua Zhou include Michigan State University & Tokyo Institute of Technology.
Papers
More filters
Journal ArticleDOI
Towards Making Unlabeled Data Never Hurt
Yu-Feng Li,Zhi-Hua Zhou +1 more
TL;DR: In this article, a safe semi-supervised support vector machine (S4VM) is proposed to improve the safety of S3VM by using multiple low-density separators to approximate the ground-truth decision boundary.
Journal ArticleDOI
Constraint Score: A new filter method for feature selection with pairwise constraints
TL;DR: This paper proposes to use another form of supervision information for feature selection, i.e. pairwise constraints, which specifies whether a pair of data samples belong to the same class (must-link constraints) or different classes (cannot- link constraints).
Journal ArticleDOI
Face recognition with one training image per person
Jianxin Wu,Zhi-Hua Zhou +1 more
TL;DR: In this paper, an extension of the eigenface technique, i.e. projection-combined principal component analysis, (PC)2A, is proposed and it requires less computational cost and achieves 3-5% higher accuracy on a gray-level frontal view face database where each person has only one training image.
Journal ArticleDOI
Image Retargeting Using Mesh Parametrization
TL;DR: This paper associates image saliency into the image mesh and regard image structure as constraints for mesh parametrization to emphasize salient objects and minimize visual distortion in image retargeting.
Proceedings Article
Multi-label learning with weak label
Yuyin Sun,Yin Zhang,Zhi-Hua Zhou +2 more
TL;DR: The WELL (WEak Label Learning) method is proposed, which considers that the classification boundary for each label should go across low density regions, and that each label generally has much smaller number of positive examples than negative examples.