Research map / Optimization / Implicit bias and generalization
Learning Compact Neural Networks with Regularization
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 |
|---|---|---|---|
| Add sparsity regularization with ℓ1 or ℓ0 projection | Shallow network trained with good initialization · Synthetic sparse-network experiment | normalized test loss: ≈0 test loss when n≥600 | Helped |
| Add ℓ1 projection instead of ℓ0 projection | Shallow network trained with good initialization · Synthetic sparse-network experiment | normalized test loss | Helped |
| Add sparsity regularization | Shallow network trained with random initialization · Synthetic sparse-network experiment | normalized test loss | Helped |
| Use ℓ1 projection instead of ℓ0 projection | Shallow network trained with random initialization · Synthetic sparse-network experiment | normalized test loss: approaches 0 only at n=2000 | Helped |
| Use random initialization instead of initialization near the ground truth | Sparsity-regularized shallow network · Synthetic sparse-network experiment | normalized test loss: n≥600; n=2000 | Hurt |
| Add convolutional weight sharing | Shallow network trained with random initialization · Synthetic CNN experiment | test loss | Helped |
| Initialize near the ground truth instead of randomly | CNN subspace-constrained model · Synthetic CNN experiment | test loss: ≈ 0 | Helped |
Closest papers by meaning
- Penalizing Gradient Norm for Efficiently Improving Generalization in Deep Learning (ICML 2022 · similarity 0.90)
- Learning Sparse Neural Networks through L_0 Regularization (ICLR 2018 · similarity 0.90)
- Global Minimizers of $\ell^p$-Regularized Objectives Yield the Sparsest ReLU Neural Networks (NeurIPS 2025 · similarity 0.88)
- Deep Weight Factorization: Sparse Learning Through the Lens of Artificial Symmetries (ICLR 2025 · similarity 0.88)
- Improving Deep Learning Optimization through Constrained Parameter Regularization (NeurIPS 2024 · similarity 0.88)
- Random Sparse Lifts: Construction, Analysis and Convergence of finite sparse networks (ICLR 2024 · 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.