Unsupervised Ensemble Learning Via Network Generation

Zhongfan Zhang, Wenming Cao, Cheng Liu, Rui Li, Qianfen Jiao, Zhiwen Yu, C. L. Philip Chen, Hau-San Wong

Abstract

In this work, we propose an unsupervised ensemble learning method via network generation, referred to as UELNG. Specifically, we first generate weights of clustering ensemble models by adopting HyperGAN, and obtain diverse partitions for data. With these partitions, we can easily identify high-confident pseudo-labels as supervised information to predict low-entropy labels for unlabeled augmented data, thereby enhancing the quality of pseudo-labels and clustering accuracy. We conduct experiments on multiple data sets. Experimental results indicate that our method outperforms state-of-the-art methods by 0.3%, 1.8%, 5.7%, 3.2% and 2.4% on MNIST, STL-10, CIFAR-10, Reuters and 20News, respectively, which demonstrates the effectiveness of our proposed UELNG.
BibTeX
@inproceedings{zhang2021unsupervised,
  author = {Zhongfan Zhang and Wenming Cao and Cheng Liu and Rui Li and Qianfen Jiao and Zhiwen Yu and C. L. Philip Chen and Hau-San Wong},
  title = {{Unsupervised Ensemble Learning Via Network Generation}},
  booktitle = {ICME},
  year = {2021},
  doi = {10.1109/icme51207.2021.9428455},
}

Loading related content...

Figure