Discriminatively boosted image clustering with fully convolutional auto-encoders
作者：
Highlights：
• Fully convolutional deep auto-encoders are proposed for image feature extraction.
• The auto-encoder can be trained in an end-to-end manner.
• A unified clustering method is constructed based on the encoder and soft k-means.
• Performance of the method can be greatly boosted in self-paced learning procedures.
• The methods can achieve state-of-the-art performance on several vision datasets.
摘要
•Fully convolutional deep auto-encoders are proposed for image feature extraction.•The auto-encoder can be trained in an end-to-end manner.•A unified clustering method is constructed based on the encoder and soft k-means.•Performance of the method can be greatly boosted in self-paced learning procedures.•The methods can achieve state-of-the-art performance on several vision datasets.
论文关键词：Image clustering,Fully convolutional auto-encoder,Representation learning,Discriminatively boosted clustering
论文评审过程：Received 23 March 2017, Revised 10 January 2018, Accepted 20 May 2018, Available online 21 May 2018, Version of Record 4 June 2018.
论文官网地址：https://doi.org/10.1016/j.patcog.2018.05.019
