Research map / Graph learning / Graph algorithms and clustering
Euler Characteristic Tools for Topological Data Analysis
Olympio Hacquard, Vadim Lebovici
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 |
|---|---|---|---|
| use random forest instead of SVR on Euler curves | ECC · Curvature regression | R2 score: 0.70 -> 0.93 | Improved |
| add hybrid transform on top of Euler curve with SVR | ECC+SVR · Curvature regression | R2 score: 0.70 -> 0.79 | Improved |
| add hybrid transform on top of Euler curve with random forest | ECC+RF · Curvature regression | R2 score: 0.93 -> 0.89 | Hurt |
| use two-parameter Euler surface instead of one-parameter Euler curve | XGB · ORBIT5K | accuracy: 83.8 -> 91.8 | Improved |
| use two-parameter hybrid transform instead of one-parameter | XGB · ORBIT5K | accuracy: 82.8 -> 89.9 | Improved |
| add hybrid transform to Euler surface | ECS + XGB · ORBIT5K | accuracy: 91.8 -> 89.9 | Hurt |
| reduce feature vector to three transform evaluation points (ablation) | hybrid transform with SVM · ORBIT5K | accuracy: approximately 80 % | No clear effect |
| use hybrid transform instead of Euler curve for compact features | hybrid transforms · ORBIT5K computation time | time (s): 16 -> 5 (one-parameter ECC to HT1); 135 -> 69 (ECS to HT2) | Improved |
| use Euler curve instead of persistence image at fixed resolution | ECC · ORBIT5K computation time | time (s): 139 -> 16 | Improved |
| use hybrid transform instead of persistence image in two-parameter case | HT2 · ORBIT5K computation time | time (s): 2034 -> 805 | Improved |
| use multi-dimensional hybrid transform instead of one-dimensional | HT · Graph classification | accuracy: 87.4 -> 89.4 on MUTAG; 73.3 -> 75.4 on PROTEINS | Improved |
Closest papers by meaning
- Differentiable Euler Characteristic Transforms for Shape Classification (ICLR 2024 · similarity 0.88)
- On topological descriptors for graph products (NeurIPS 2025 · similarity 0.87)
- Diss-l-ECT: Dissecting Graph Data with Local Euler Characteristic Transforms (ICML 2025 · similarity 0.85)
- Point-Level Topological Representation Learning on Point Clouds (ICML 2025 · similarity 0.85)
- Diffeomorphic interpolation for efficient persistence-based topological optimization (NeurIPS 2024 · similarity 0.83)
- Differentiable Mapper for Topological Optimization of Data Representation (ICML 2024 · similarity 0.82)
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.
