Research map / Adversarial robustness and security / Black-box and transfer attacks
An Efficient Adversarial Attack for Tree Ensembles
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 |
|---|---|---|---|
| restrict neighborhood to bound trees at hamming distance 1 (LT-Attack) instead of enumerating all leaves (NaiveLeaf) (ablation) | LT-Attack · MNIST standard GBDT, l2 | time per example: 2.37s -> .105s (22.6X) | Helped |
| restrict neighborhood to bound trees instead of enumerating all leaves (NaiveLeaf) (ablation) | LT-Attack · MNIST standard GBDT, l2 | l2 perturbation (lower better): .081 -> .108 | Hurt |
| use leaf-tuple hamming-1 neighborhood instead of one-feature-at-a-time mutation (NaiveFeature) (ablation) | LT-Attack · MNIST standard GBDT, l2 | l2 perturbation (lower better): .229 -> .108 | Helped |
| use leaf-tuple hamming-1 neighborhood instead of one-feature-at-a-time mutation (ablation) | LT-Attack · F-MNIST standard GBDT, l2 | l2 perturbation (lower better): .181 -> .096 | Helped |
| LT-Attack instead of Cube | tree ensemble attack · MNIST standard GBDT, l_inf | avg l_inf perturbation (lower better): .056 -> .029 | Helped |
| LT-Attack instead of SignOPT/HSJA decision-based attacks | tree ensemble attack · MNIST standard GBDT, l_inf | avg l_inf perturbation (lower better): .195 -> .029 vs SignOPT; .117 -> .029 vs HSJA | Helped |
| LT-Attack on robustly trained GBDT instead of Cube | tree ensemble attack · MNIST robust GBDT, l_inf | avg l_inf perturbation (lower better): .296 -> .290 | Helped |
Closest papers by meaning
- Faster Repeated Evasion Attacks in Tree Ensembles (NeurIPS 2024 · similarity 0.91)
- Evasion and Hardening of Tree Ensemble Classifiers (ICML 2016 · similarity 0.89)
- Robustness Verification of Multi-Class Tree Ensembles (AAAI 2024 · similarity 0.87)
- TTTS: Tree Test Time Simulation for Enhancing Decision Tree Robustness against Adversarial Examples (AAAI 2024 · similarity 0.86)
- Learning Decision Trees and Forests with Algorithmic Recourse (ICML 2024 · similarity 0.85)
- Cultivating Archipelago of Forests: Evolving Robust Decision Trees Through Island Coevolution (AAAI 2025 · similarity 0.83)
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.