Research map / Learning theory and algorithms / Sample complexity
ReLU Regression with Massart Noise
Ilias Diakonikolas, Jongho Park, Christos Tzamos
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 |
|---|---|---|---|
| add radial-isotropic transformation to l1 regression | l1 regression · Synthetic linear, eta varied, 120 samples | exact recovery rate | Improved |
| add radial-isotropic transformation to l1 regression | l1 regression · Synthetic linear, eta=0.25, sample size varied | exact recovery rate | Improved |
| add normalization preprocessing to l1 regression | l1 regression · Synthetic linear mixture of Gaussians | exact recovery rate | No clear effect |
| add isotropic transformation preprocessing to l1 regression | l1 regression · Synthetic linear | exact recovery rate | No clear effect |
| apply radial-isotropic transformation inside gradient descent on l1 loss | gradient descent · Synthetic ReLU, eta=0.4, 240 samples | l2 distance to optimum | Improved |
| apply isotropic transformation inside gradient descent on l1 loss | gradient descent · Synthetic ReLU, eta=0.4 | l2 distance to optimum | Mixed |
| use Rescaled L1 instead of naive l1 regression at minimal noise | l1 regression · Drug discovery dataset | fraction of test points within margin | Hurt |
| use Rescaled L1 instead of baselines at moderate to high noise | l1 regression, least squares, ridge · Drug discovery dataset | fraction of test points within margin | Improved |
Closest papers by meaning
- Finite-Sample Analysis of Learning High-Dimensional Single ReLU Neuron (ICML 2023 · similarity 0.87)
- Benign overfitting in leaky ReLU networks with moderate input dimension (NeurIPS 2024 · similarity 0.86)
- Stable Minima Cannot Overfit in Univariate ReLU Networks: Generalization by Large Step Sizes (NeurIPS 2024 · similarity 0.86)
- Noisy Interpolation Learning with Shallow Univariate ReLU Networks (ICLR 2024 · similarity 0.85)
- Online Linear Classification with Massart Noise (ICML 2025 · similarity 0.85)
- Robust Regression of General ReLUs with Queries (NeurIPS 2025 · 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.
