Factorized Convolutional Networks: Unsupervised Fine-Tuning for Image Clustering
2018 IEEE WINTER CONFERENCE ON APPLICATIONS OF COMPUTER VISION (WACV 2018)(2018)
摘要
Deep convolutional neural networks (CNNs) have recognized promise as universal representations for various image recognition tasks. One of their properties is the ability to transfer knowledge from a large annotated source dataset (e.g., ImageNet) to a (typically smaller) target dataset. This is usually accomplished through supervised fine-tuning on labeled new target data. In this work, we address "unsupervised fine-tuning" that transfers a pre-trained network to target tasks with unlabeled data such as image clustering tasks. To this end, we introduce group-sparse non-negative matrix factorization (GSNMF), a variant of NMF, to identify a rich set of high-level latent variables that are informative on the target task. The resulting "factorized convolutional network" (FCN) can itself be seen as a feed-forward model that combines CNN and two-layer structured NMF. We empirically validate our approach and demonstrate state-of-the-art image clustering performance on challenging scene (MIT-67) and fine-grained (Birds-200, Flowers-102) benchmarks. We further show that, when used as unsupervised initialization, our approach improves image classification performance as well.
更多查看译文
关键词
deep convolutional neural networks,fine-grained benchmarks,group-sparse nonnegative matrix factorization,feed-forward model,two-layer structured NMF,CNNs,GSNMF,image classification performance,unsupervised initialization,high-level latent variables,image clustering tasks,supervised fine-tuning,target dataset,image recognition tasks,universal representations,unsupervised fine-tuning,factorized convolutional networks
AI 理解论文
溯源树
样例
生成溯源树,研究论文发展脉络
Chat Paper
正在生成论文摘要