Research map / Optimization / Implicit bias and generalization
Principal Components Bias in Over-parameterized Linear Models, and its Manifestation in Deep Neural Networks
Guy Hacohen, Daphna Weinshall
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 |
|---|---|---|---|
| whiten the data (ZCA) to remove PC-bias (ablation) | two-layer linear networks · cats and dogs | learning order consistency (accessibility correlation) | Hurt |
| whiten the data (ZCA) to remove PC-bias (ablation) | VGG-19 on CIFAR-10 · CIFAR-10 | learning order consistency: down by half | Mixed |
| whiten images before training with shuffled labels (ablation) | two-layer linear networks · natural images | convergence rate | Improved |
| amplify lowest principal components of CIFAR-10 (ablation) | st-VGG and linear st-VGG · CIFAR-10 | benefit of early stopping | Improved |
| evaluate on test set projected to few principal components (P=1,10) (ablation) | st-VGG · small mammals | test accuracy over epochs | Mixed |
Closest papers by meaning
- Bias of Stochastic Gradient Descent or the Architecture: Disentangling the Effects of Overparameterization of Neural Networks (ICML 2024 · similarity 0.87)
- Simplicity Bias in Overparameterized Machine Learning (AAAI 2024 · similarity 0.86)
- Attention-based PCA (NeurIPS 2026 · similarity 0.86)
- Simplicity Bias and Optimization Threshold in Two-Layer ReLU Networks (ICML 2025 · similarity 0.85)
- Initial Guessing Bias: How Untrained Networks Favor Some Classes (ICML 2024 · similarity 0.85)
- Tilting the Odds at the Lottery: the Interplay of Overparameterisation and Curricula in Neural Networks (ICML 2024 · similarity 0.85)
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.
