Research map / Learning theory and algorithms / Sample complexity
L1-regularized Neural Networks are Improperly Learnable in Polynomial Time
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 |
|---|---|---|---|
| increase depth from k=1 to k=4 in the recursive kernel method (ablation) | recursive kernel method · MNIST basic and variations | classification error rate: small gains on all four datasets (e.g., Basic 3.31% to 3.08%) | Helped |
| use recursive kernel method instead of multilayer perceptron | MNIST variations · MNIST Basic, Rotation, Background, Background+Rotation | classification error rate: consistently lower error than MLP on all four datasets | Helped |
| use recursive kernel method instead of LeNet5 convolutional network | MNIST variations · MNIST Background and Background+Rotation | classification error rate: 22.13% vs 9.35% on Background; 52.94% vs 32.36% on Background+Rotation | Hurt |
| recursive kernel method (k=4) vs LeNet5 on rotated digits | MNIST Rotation · MNIST Rotation | classification error rate: 8.78% vs 9.27%, best of all methods on Rotation | Mixed |
Closest papers by meaning
- Learning Sparse Neural Networks through L_0 Regularization (ICLR 2018 · similarity 0.89)
- Generalization Bounds for Neural Networks via Approximate Description Length (NeurIPS 2019 · similarity 0.87)
- The Sample Complexity of One-Hidden-Layer Neural Networks (NeurIPS 2022 · similarity 0.87)
- Complexity of Neural Network Training and ETR: Extensions with Effectively Continuous Functions (AAAI 2024 · similarity 0.87)
- A New Neural Kernel Regime: The Inductive Bias of Multi-Task Learning (NeurIPS 2024 · similarity 0.86)
- Global Minimizers of $\ell^p$-Regularized Objectives Yield the Sparsest ReLU Neural Networks (NeurIPS 2025 · 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.