Research map / Label-efficient and robust learning / Recognition and long-tail
Infinite Feature Selection: A Graph-based Feature Filtering Approach
Giorgio Roffo, Simone Melzi, Umberto Castellani, Alessandro Vinciarelli, Marco Cristani
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 class labels (supervised Inf-FS_S instead of Inf-FS_U) | Inf-FS · small-sample microarray datasets | accuracy | Improved |
| use class labels (Inf-FS_S instead of Inf-FS_U) on DEXTER | Inf-FS · DEXTER (automatic subset) | accuracy (%): 83.8 -> 92.8 | Improved |
| use class labels (Inf-FS_S instead of Inf-FS_U) on GISETTE | Inf-FS · GISETTE (automatic subset) | accuracy (%): 87.7 -> 94.1 | Improved |
| use class labels (Inf-FS_S instead of Inf-FS_U) on GINA | Inf-FS · GINA (automatic subset) | accuracy (%): 76.2 -> 83.4 | Improved |
| use class labels (Inf-FS_S instead of Inf-FS_U) on MADELON | Inf-FS · MADELON (automatic subset) | accuracy (%): 58.7 -> 57.1 | Hurt |
| use class labels (Inf-FS_S instead of Inf-FS_U) on LYMPHOMA | Inf-FS · LYMPHOMA (automatic subset) | accuracy (%): 93.3 -> 95.8 | Improved |
| automatic subset selection via clustering with Inf-FS instead of LASSO | Inf-FS_U / Inf-FS_S · 11 benchmarks | accuracy | Mixed |
| supervised Inf-FS_S instead of unsupervised for automatic subset size | Inf-FS · all datasets | number of features kept | Improved |
| keep top 25% of CNN features with Inf-FS_S instead of the full feature set | Inf-FS_S · PASCAL VOC 2007 | mAP (%): 83.1 -> 83.8 | Improved |
| fine-grained cardinality search (31.5% of features) for Inf-FS_S | Inf-FS_S · PASCAL VOC 2007 | mAP (%): 83.8 -> 84.18 | Improved |
| fine-grained cardinality search (36.5% of features) for Inf-FS_U | Inf-FS_U · PASCAL VOC 2007 | mAP (%): 83.6 -> 83.91 | Improved |
| retain the last-ranked 10% of features (90-100% range) with Inf-FS_S | Inf-FS_S · PASCAL VOC 2007 | mAP | Hurt |
Closest papers by meaning
- Infinite Latent Feature Selection: A Probabilistic Latent Graph-Based Ranking Approach (ICCV 2017 · similarity 0.87)
- Adaptive Node Feature Selection for Graph Neural Networks (ICML 2026 · similarity 0.87)
- HeroFilter: Adaptive Spectral Graph Filter for Varying Heterophilic Relations (NeurIPS 2025 · similarity 0.85)
- Homophily-Related: Adaptive Hybrid Graph Filter for Multi-View Graph Clustering (AAAI 2024 · similarity 0.84)
- Embedded Feature Selection on Graph-Based Multi-View Clustering (AAAI 2024 · similarity 0.84)
- Hypernetworks for Dynamic Feature Selection (NeurIPS 2026 · 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.
