端到端优化谱聚类，用最优传输提升聚类精度
Bootstrap Deep Spectral Clustering with Optimal Transport
- 用单一网络联合优化谱聚类全流程
- 在ImageNet-Dogs上比次优方法高16% NMI
- 适合需要高精度聚类的视觉任务
谱聚类是一种主流聚类方法，但存在优化过程分离和表征能力有限两大缺陷。为此，我们提出一种深度谱聚类模型BootSC，以端到端方式联合学习谱聚类的所有阶段——亲和矩阵构建、谱嵌入和k-means聚类。BootSC利用高效且有效的最优传输导出的监督信号来引导亲和矩阵和聚类分配矩阵的生成。此外，引入语义一致的正交重参数化技术，对谱嵌入进行正交化，显著增强其判别能力。实验结果表明，BootSC达到当前最优聚类性能，例如在具有挑战性的ImageNet-Dogs数据集上，相比次优方法实现了16%的NMI提升。代码已公开于https://github.com/spdj2271/BootSC。
原文摘要 · Abstract (English)
Spectral clustering is a leading clustering method. Two of its major shortcomings are the disjoint optimization process and the limited representation capacity. To address these issues, we propose a deep spectral clustering model (named BootSC), which jointly learns all stages of spectral clustering -- affinity matrix construction, spectral embedding, and $k$-means clustering -- using a single network in an end-to-end manner. BootSC leverages effective and efficient optimal-transport-derived supervision to bootstrap the affinity matrix and the cluster assignment matrix. Moreover, a semantically-consistent orthogonal re-parameterization technique is introduced to orthogonalize spectral embeddings, significantly enhancing the discrimination capability. Experimental results indicate that BootSC achieves state-of-the-art clustering performance. For example, it accomplishes a notable 16\% NMI improvement over the runner-up method on the challenging ImageNet-Dogs dataset. Our code is available at https://github.com/spdj2271/BootSC.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。