Research map / Learning theory and algorithms / Prediction and decision losses
Adversarial Multiclass Classification: A Risk Minimization Perspective
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 |
|---|---|---|---|
| Recast zero-one adversarial classification as ERM with the AL0-1 loss (vs. convex hinge surrogates) | multiclass zero-one adversarial classifier · UCI 12 datasets | average accuracy (linear kernel): 81.59 vs 81.25 (CS), 81.02 (WW), 68.80 (LLW) | Helped |
| Use Gaussian RBF kernel instead of linear features | AL0-1 · UCI 12 datasets | average accuracy: 85.14 vs 81.59 (linear) | Helped |
| Absolute-margin LLW loss vs. relative-margin losses with linear features | LLW multiclass SVM · UCI 12 datasets, linear kernel | average accuracy: 68.80 vs ~81 for AL0-1, WW, CS | Hurt |
| Constraint generation instead of enumerating all 2^|Y|-1 constraints (ablation) | AL0-1 dual QP · optdigits (Gaussian kernel) | constraints added to working set: 10072 added and 2315 active vs. ~3.9 million possible constraints | Helped |
| Gaussian kernel vs linear for LLW | LLW · UCI 12 datasets | average accuracy: 84.93 vs 68.80 | Helped |
Closest papers by meaning
- Theoretical Analysis of Adversarial Learning: A Minimax Approach (NeurIPS 2019 · similarity 0.90)
- Towards Consistency in Adversarial Classification (NeurIPS 2022 · similarity 0.88)
- Adversarial Training Can Provably Improve Robustness: Theoretical Analysis of Feature Learning Process Under Structured Data (ICLR 2025 · similarity 0.88)
- Adversarial Training Should Be Cast as a Non-Zero-Sum Game (ICLR 2024 · similarity 0.87)
- Kernel Learning with Adversarial Features: Numerical Efficiency and Adaptive Regularization (NeurIPS 2025 · similarity 0.87)
- Consistent Adversarially Robust Linear Classification: Non-Parametric Setting (ICML 2024 · 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.