Research map / Detection and segmentation / Open-set and few-shot detection
Detecting Interactions from Neural Networks via Topological Analysis
Zirui Liu, Qingquan Song, Kaixiong Zhou, Ting-Hsiang Wang, Ying Shan, Xia Hu
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 PID instead of NID for pairwise interaction detection | MLP interaction detector · Synthetic F1-F10 average | AUC: 0.951±7.0e−2 -> 0.957±6.2e−2 | Improved |
| Use PID instead of NID on F2 | MLP interaction detector · Synthetic F2 | AUC: 0.776±4.3e−2 -> 0.804±5.7e-2 | Improved |
| Use PID instead of NID on F4 | MLP interaction detector · Synthetic F4 | AUC: 0.916±6.3e−2 -> 0.935±3.9e−2 | Improved |
| Use PID instead of NID on F1 | MLP interaction detector · Synthetic F1 | AUC: 0.985±6.3e−3 -> 0.986±4.1e−3 | No clear effect |
| Use PID instead of NID on F10 | MLP interaction detector · Synthetic F10 | AUC: 0.989±3.0e−2 -> 0.987±3.5e−2 | No clear effect |
| Use PID instead of tree-based Additive Groves on F5 | Interaction detector · Synthetic F5 | AUC: 0.67±5.7e-2 -> 1.0±0.0 | Improved |
| Use PID instead of tree-based Additive Groves on F6 | Interaction detector · Synthetic F6 | AUC: 0.64±1.4e-2 -> 1.0±0.0 | Improved |
| Use PID instead of Additive Groves on F10 | Interaction detector · Synthetic F10 | AUC: 1.0±0.0 -> 0.987±3.5e−2 | No clear effect |
| Add PID-generated synthetic features to random forest | Random forest · Amazon Employee | AUC: 0.8378±0.0046 -> 0.8460±0.0079 | Improved |
| Add NID-generated synthetic features to random forest | Random forest · Amazon Employee | AUC: 0.8378±0.0046 -> 0.8321±0.0299 | No clear effect |
| Add random synthetic features to random forest | Random forest · Amazon Employee | AUC: 0.8378±0.0046 -> 0.7780±0.0575 | Hurt |
| Add PID features to random forest | Random forest · Higgs Boson | AUC: 0.7421±0.0019 -> 0.7422±0.0017 | No clear effect |
Closest papers by meaning
- TopoGCL: Topological Graph Contrastive Learning (AAAI 2024 · similarity 0.85)
- Graphcode: Learning from multiparameter persistent homology using graph neural networks (NeurIPS 2024 · similarity 0.85)
- Boosting Graph Pooling with Persistent Homology (NeurIPS 2024 · similarity 0.85)
- TopInG: Topologically Interpretable Graph Learning via Persistent Rationale Filtration (ICML 2025 · similarity 0.84)
- Neural Persistence Dynamics (NeurIPS 2024 · similarity 0.84)
- Persistent Topological Features in Large Language Models (ICML 2025 · similarity 0.84)
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.
