Research map / Label-efficient and robust learning / Recognition and long-tail
Neural Representations Reveal Distinct Modes of Class Fitting in Residual Convolutional Networks
Michał Jamroż, Marcin Kurdziel
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (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 |
|---|---|---|---|
| Compare high-density classes with low-density classes | ResNet class representations · CIFAR-100 and Mini-ImageNet | representation-distribution complexity | Improved |
| Compare high-density classes with low-density classes | ResNet class representations · CIFAR-100 and Mini-ImageNet | input memorization | Mixed |
| Compare high-density classes with low-density classes | Smoothed ResNet classifiers · CIFAR-100 and Mini-ImageNet | certified radius | Hurt |
| Compare high-density classes with low-density classes | Smoothed ResNet classifiers · CIFAR-100 and Mini-ImageNet | F-score | Hurt |
| Compare high-density classes with low-density classes | Smoothed ResNet classifiers · CIFAR-100 and Mini-ImageNet | classification rate | Hurt |
| Compare memorized examples with typical examples | ResNet class representations · CIFAR-100 and Mini-ImageNet | between-class representation divergence | Hurt |
| Compare early stages with the third and fourth stages | ResNet · CIFAR-100 and Mini-ImageNet | distinct class-fitting modes | Improved |
| Replace the residual architecture with a plain 8-layer convolutional network | Convolutional network class representations · CIFAR-100 and Mini-ImageNet | distinct class-fitting modes | Hurt |
Closest papers by meaning
- Understanding Robust Learning through the Lens of Representation Similarities (NeurIPS 2022 · similarity 0.88)
- Training Interpretable Convolutional Neural Networks by Differentiating Class-specific Filters (ECCV 2020 · similarity 0.87)
- Network Dissection: Quantifying Interpretability of Deep Visual Representations (CVPR 2017 · similarity 0.87)
- Interpreting Deep Visual Representations via Network Dissection (TPAMI 2018 · similarity 0.86)
- Tracing Representation Progression: Analyzing and Enhancing Layer-Wise Similarity (ICLR 2025 · similarity 0.86)
- When Representations Align: Universality in Representation Learning Dynamics (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.
