Research map / Label-efficient and robust learning / Recognition and long-tail
Utilizing Class Information for Deep Network Representation Shaping
Daeyoung Choi, Wonjong Rhee
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 |
|---|---|---|---|
| Apply class-wise variance regularization | CNN · CIFAR-10, Momentum | error (%): 25.78 ± 0.37 -> 24.42 ± 0.31 | Improved |
| Apply class-wise variance regularization | ResNet-32 · CIFAR-10 | error (%): 7.39 -> 7.17 | Improved |
| Apply class-wise variance regularization | ResNet-110 · CIFAR-10 | error (%): 6.12 -> 6.00 | Improved |
| Apply class-wise variance regularization | 6-layer autoencoder · MNIST autoencoder | mean squared error: 1.44 × 10^{-2} ± 3.36 × 10^{-4} -> 1.19 × 10^{-2} ± 2.48 × 10^{-4} | Improved |
| Use class-wise covariance regularization instead of class-wise variance regularization (ablation) | 6-layer MLP · MNIST MLP, default | error (%): 2.42 ± 0.06 -> 2.49 ± 0.10 | Hurt |
| Remove class conditioning from variance regularization (ablation) | 6-layer MLP · MNIST MLP, default | error (%): 2.42 ± 0.06 -> 2.65 ± 0.11 | Hurt |
| Apply all-class variance regularization rather than no regularizer (ablation) | CNN · CIFAR-10, Adam | error (%): 26.64 ± 0.16 -> 21.44 ± 0.88 | Improved |
| Move representation regularization from lower layers toward upper layers (ablation) | deep networks · MNIST MLP and CIFAR-10 CNN | error (%) | Mixed |
Closest papers by meaning
- Network-to-Network Regularization: Enforcing Occam's Razor to Improve Generalization (NeurIPS 2021 · similarity 0.86)
- T-vMF Similarity for Regularizing Intra-Class Feature Distribution (CVPR 2021 · similarity 0.86)
- Reducing Class-Wise Performance Disparity via Margin Regularization (ICLR 2026 · similarity 0.86)
- Neural Representations Reveal Distinct Modes of Class Fitting in Residual Convolutional Networks (AAAI 2023 · similarity 0.85)
- Graph Your Own Prompt (NeurIPS 2025 · similarity 0.84)
- Learning from Teaching Regularization: Generalizable Correlations Should be Easy to Imitate (NeurIPS 2024 · similarity 0.84)
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.
