Research map / Learning theory and algorithms / Combinatorial and search optimization
Robustness Verification of Tree-based Models
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 the multilevel clique verifier instead of MILP | GBDT robustness verification · Natural and robust GBDTs across the nine reported datasets | verification time: up to ∼ 3000X speedup | Helped |
| Use the multilevel clique verifier instead of LP relaxation | GBDT robustness verification · Natural and robust GBDTs | lower-bound quality | Helped |
| Increase the number of clique-enumeration levels (ablation) | Multilevel verifier · 20-tree GBDT on diabetes | lower-bound tightness | Helped |
| Use larger cliques at each level (ablation) | Multilevel verifier · 20-tree GBDT on diabetes | lower-bound tightness | Helped |
| Add dynamic programming for the final single-layer bound (ablation) | Multilevel verifier · Natural and robust GBDTs across datasets | lower-bound distortion: .208 -> .209 (breast-cancer natural); .109 -> .111 (diabetes robust); .026 -> .027 (ijcnn1 natural) | Helped |
Closest papers by meaning
- Versatile Verification of Tree Ensembles (ICML 2021 · similarity 0.93)
- Robustness Verification of Multi-Class Tree Ensembles (AAAI 2024 · similarity 0.92)
- Abstract Interpretation of Decision Tree Ensemble Classifiers (AAAI 2020 · similarity 0.92)
- Sensitivity Verification for Additive Decision Tree Ensembles (ICLR 2025 · similarity 0.92)
- TTTS: Tree Test Time Simulation for Enhancing Decision Tree Robustness against Adversarial Examples (AAAI 2024 · similarity 0.88)
- Probably Approximately Global Robustness Certification (ICML 2025 · 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.