Research map / Learning theory and algorithms / Combinatorial and search optimization
GenSVM: A Generalized Multiclass Support Vector Machine
Gerrit J.J. van den Burg, Patrick J.F. Groenen
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 GenSVM instead of MSVMpack Lee et al. (LLW) multiclass SVM | GenSVM · Small data sets (nested CV) | ARI rank: p < 10-14 | Improved |
| use GenSVM instead of MSVMpack MSVM2 (Guermeur and Monfrini) | GenSVM · Small data sets (nested CV) | ARI rank: p = 10-6 | Improved |
| use GenSVM instead of MSVMpack Crammer and Singer | GenSVM · Small data sets (nested CV) | ARI rank: p = 4 · 10-5 | Improved |
| use GenSVM instead of LibLinear Crammer and Singer | GenSVM · Small data sets (nested CV) | ARI rank: p = 0.0004 | Improved |
| use warm starts in GenSVM grid search (GenSVM vs other methods, average time per configuration) | GenSVM · Small data sets (nested CV, grid search) | average training time per configuration (s): 0.97 seconds vs 20.56 seconds for LibLinear C&S, 24.84 seconds for OvO, 25.03 seconds for DAGSVM | Improved |
| use GenSVM despite a grid 18 times larger | GenSVM · Small data sets (nested CV, grid search) | probability of fastest total grid search: 40.35% vs 22.81% for OvO, 19.30% for DAG, 17.54% for LibLinear C&S | Improved |
| use one-vs-one instead of GenSVM for best classification probability | GenSVM · Small data sets (nested CV) | probability of best performance Pc(1): 28.42% for OvO vs 26.32% for GenSVM and DAGSVM | No clear effect |
| use one-vs-all instead of OvO, DAG or GenSVM | OvA · Small data sets (nested CV) | ARI | Hurt |
| GenSVM vs LibLinear C&S on Covtype | GenSVM · Covtype (large) | ARI: 0.3571 vs 0.3582 | No clear effect |
| GenSVM vs other methods on Fars | GenSVM · Fars (large) | ARI: 0.8102 vs 0.8090 | Improved |
| GenSVM vs LibSVM DAG on KDDCup-99 | GenSVM · KDDCup-99 (large) | ARI: 0.9758 vs 0.9809 | Hurt |
Closest papers by meaning
- A Unified View on Multi-class Support Vector Classification (JMLR 2016 · similarity 0.89)
- Fast Support Vector Classification for Large-Scale Problems (TPAMI 2021 · similarity 0.87)
- Multiclass Anomaly Detector: the CS++ Support Vector Machine (JMLR 2020 · similarity 0.86)
- Equivalence of Multicategory SVM and Simplex Cone SVM: Fast Computations and Statistical Theory (ICML 2018 · similarity 0.86)
- Multi-Class Support Vector Machine with Maximizing Minimum Margin (AAAI 2024 · similarity 0.86)
- ThunderSVM: A Fast SVM Library on GPUs and CPUs (JMLR 2018 · similarity 0.86)
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.
