Research map / Learning theory and algorithms / Prediction and decision losses
On the Error Resistance of Hinge-Loss Minimization
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 |
|---|---|---|---|
| Use hinge-loss minimization under the dense pancakes and HerSumNorm conditions | Hinge-loss minimization · Margin-separable inlier distribution with adversarial outliers | accuracy: accuracy at least (1 – β) | Helped |
| Allow adversarial label flips while keeping feature points unchanged | Surrogate loss minimization · Uniform mixture of O(1) logconcave distributions | accuracy: η = O(min(1/3, γ/2µ, γ/(√d log(γ√d))); accuracy 1 – β | Helped |
| Require margin at least C log 1/β / √d | Surrogate loss minimization · Isotropic logconcave class-conditional distributions | accuracy: γ ≥ C log 1/β/√d; accuracy 1 – β | Helped |
Closest papers by meaning
- Trading off Consistency and Dimensionality of Convex Surrogates for Multiclass Classification (NeurIPS 2024 · similarity 0.88)
- H-Consistency Bounds for Surrogate Loss Minimizers (ICML 2022 · similarity 0.88)
- Toward Minimal-dimensional Convex Calibrated Surrogate Losses for Classification with Rejection (NeurIPS 2026 · similarity 0.88)
- Calibration and Consistency of Adversarial Surrogate Losses (NeurIPS 2021 · similarity 0.88)
- A Universal Growth Rate for Learning with Smooth Surrogate Losses (NeurIPS 2024 · similarity 0.87)
- The Pessimistic Limits and Possibilities of Margin-based Losses in Semi-supervised Learning (NeurIPS 2018 · similarity 0.87)
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.