Scaling Up Deep Clustering Methods Beyond ImageNet-1K
arXiv preprint arXiv:2406.01203, 2024•arxiv.org
Deep image clustering methods are typically evaluated on small-scale balanced
classification datasets while feature-based $ k $-means has been applied on proprietary
billion-scale datasets. In this work, we explore the performance of feature-based deep
clustering approaches on large-scale benchmarks whilst disentangling the impact of the
following data-related factors: i) class imbalance, ii) class granularity, iii) easy-to-recognize
classes, and iv) the ability to capture multiple classes. Consequently, we develop multiple …
classification datasets while feature-based $ k $-means has been applied on proprietary
billion-scale datasets. In this work, we explore the performance of feature-based deep
clustering approaches on large-scale benchmarks whilst disentangling the impact of the
following data-related factors: i) class imbalance, ii) class granularity, iii) easy-to-recognize
classes, and iv) the ability to capture multiple classes. Consequently, we develop multiple …
Deep image clustering methods are typically evaluated on small-scale balanced classification datasets while feature-based -means has been applied on proprietary billion-scale datasets. In this work, we explore the performance of feature-based deep clustering approaches on large-scale benchmarks whilst disentangling the impact of the following data-related factors: i) class imbalance, ii) class granularity, iii) easy-to-recognize classes, and iv) the ability to capture multiple classes. Consequently, we develop multiple new benchmarks based on ImageNet21K. Our experimental analysis reveals that feature-based -means is often unfairly evaluated on balanced datasets. However, deep clustering methods outperform -means across most large-scale benchmarks. Interestingly, -means underperforms on easy-to-classify benchmarks by large margins. The performance gap, however, diminishes on the highest data regimes such as ImageNet21K. Finally, we find that non-primary cluster predictions capture meaningful classes (i.e. coarser classes).
arxiv.org
Showing the best result for this search. See all results