Research map / Label-efficient and robust learning / Recognition and long-tail
Clustering-friendly Representation Learning via Instance Discrimination and Feature Decorrelation
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Add softmax feature decorrelation to instance discrimination (ablation) | IDFD · CIFAR-10 | ACC: 77.6 -> 81.5 | Helped |
| Add the orthogonality constraint to instance discrimination (ablation) | IDFO · CIFAR-10 | ACC: 77.6 -> 82.8 | Helped |
| Use instance discrimination temperature τ=1 instead of τ=0.07 (ablation) | Instance discrimination · CIFAR-10 | ACC: 44.0 -> 77.6 | Helped |
| Replace softmax decorrelation with the orthogonality constraint (ablation) | Instance discrimination with feature constraint · CIFAR-10 | ACC: 77.6 -> 82.8 | Helped |
| Use softmax decorrelation instead of the orthogonality constraint (ablation) | Instance discrimination with feature constraint · CIFAR-10 | ACC: 82.8 -> 81.5 | Mixed |
| Add feature decorrelation to the autoencoder (ablation) | AE · CIFAR-10 | ACC: 26.0 -> 22.4 | Hurt |
| Use ResNet34 instead of ResNet18 (ablation) | IDFD · CIFAR-10 | ACC: 81.5 -> 82.7 | Helped |
| Add feature decorrelation to instance discrimination (ablation) | IDFD · CIFAR-10 with ConvNet | ACC: 26.8 -> 42.0 | Helped |
| Add feature decorrelation to instance discrimination (ablation) | IDFD · CIFAR-10 with VGG16 | ACC: 39.3 -> 56.8 | Helped |
| Remove data augmentation (ablation) | IDFD · CIFAR-10 | ACC: 81.5 -> 23.6 | Hurt |
| Use data augmentation (ablation) | ID · CIFAR-10 | ACC: 18.7 -> 76.6 | Helped |
Closest papers by meaning
- Deep Spectral Clustering Learning (ICML 2017 · similarity 0.89)
- Joint Unsupervised Learning of Deep Representations and Image Clusters (CVPR 2016 · similarity 0.89)
- Deep Spectral Clustering Using Dual Autoencoder Network (CVPR 2019 · similarity 0.89)
- Unsupervised Visual Representation Learning by Online Constrained K-Means (CVPR 2022 · similarity 0.88)
- Exploring a Principled Framework for Deep Subspace Clustering (ICLR 2025 · similarity 0.86)
- Hierarchical Correlation Clustering and Tree Preserving Embedding (CVPR 2024 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.