Research map / Optimization / Implicit bias and generalization
On the Regularization Properties of Structured Dropout
Check 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 |
|---|---|---|---|
| Train with stochastic DropBlock | Single hidden-layer linear network · Synthetic linear dataset | objective equivalence | Helped |
| Train with stochastic DropConnect | Single hidden-layer linear network · Synthetic linear dataset | objective equivalence | Helped |
| Train with DropBlock | Single hidden-layer linear network · Synthetic linear dataset | convergence to theoretical global minimum | Helped |
| Apply last-layer DropBlock | ResNet-50 with an added fully connected layer · MNIST | agreement with predicted lower bound | Helped |
| Apply last-layer DropBlock | ResNet-50 with an added fully connected layer · CIFAR10 | agreement with predicted lower bound | Helped |
| Approximate original DropBlock with non-overlapping blocks (ablation) | DropBlock · Synthetic linear dataset | objective-curve agreement | No clear effect |
| Scale the original DropBlock retain probability to match the approximation's neuron-drop probability (ablation) | Original DropBlock · Synthetic linear dataset | objective-curve agreement | No clear effect |
| Apply DropConnect instead of Dropout | Single-layer linear network with squared loss · Synthetic linear dataset | expected objective | No clear effect |
Closest papers by meaning
- On Dropout and Nuclear Norm Regularization (ICML 2019 · similarity 0.91)
- The Implicit and Explicit Regularization Effects of Dropout (ICML 2020 · similarity 0.91)
- Beyond Random Masking: When Dropout meets Graph Convolutional Networks (ICLR 2025 · similarity 0.91)
- Dropout Reduces Underfitting (ICML 2023 · similarity 0.90)
- Stochastic Modified Equations and Dynamics of Dropout Algorithm (ICLR 2024 · similarity 0.90)
- Effects of Dropout on Performance in Long-range Graph Learning Tasks (NeurIPS 2025 · similarity 0.88)
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.