Research map / Label-efficient and robust learning / Recognition and long-tail
Intraclass clustering: an implicit learning ability that regularizes DNNs
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 |
|---|---|---|---|
| Use intraclass-clustering measures instead of sharpness measures to estimate generalization | Wide ResNet on CIFAR100 superclasses · CIFAR100 superclasses | granulated Kendall coefficient: 0.73 vs 0.47 | Helped |
| Use the neuron-variance measure c3 instead of sharpness measures | Wide ResNet · CIFAR100 | granulated Kendall coefficient: 0.82 vs 0.48 | Helped |
| Use the neuron-variance measure c3 instead of sharpness measures | VGG · CIFAR10 | granulated Kendall coefficient: 0.84 vs 0.68 | Helped |
| Vary the top-k neuron count in c1 and c3 (ablation) | Neuron-level intraclass-clustering measures · CIFAR100 | granulated Kendall coefficient: above 0.7 for any k in the range [1,900] | No clear effect |
| Use a single neuron per subclass for c1 (ablation) | Wide ResNet · CIFAR100 superclasses | granulated Kendall coefficient: 0.69 | Helped |
| Use a single neuron per class for c3 (ablation) | VGG · CIFAR10 | granulated Kendall coefficient: 0.78 | Helped |
| Train with data augmentation rather than without it | Wide ResNet and VGG models · CIFAR classification datasets | relationship between intraclass clustering and generalization | Helped |
Closest papers by meaning
- Understanding deep learning requires rethinking generalization (ICLR 2017 · similarity 0.87)
- Supervised autoencoders: Improving generalization performance with unsupervised regularizers (NeurIPS 2018 · similarity 0.87)
- Learning from Teaching Regularization: Generalizable Correlations Should be Easy to Imitate (NeurIPS 2024 · similarity 0.87)
- Changing the Training Data Distribution to Reduce Simplicity Bias Improves In-distribution Generalization (NeurIPS 2024 · similarity 0.86)
- Unveiling the Unseen: Identifiable Clusters in Trained Depthwise Convolutional Kernels (ICLR 2024 · similarity 0.86)
- Expand-and-Cluster: Parameter Recovery of Neural Networks (ICML 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.