Research map / Architectures and efficiency / RNNs and normalization
Approximate Vanishing Ideal via Data Knotting
Hiroshi Kera, Yoshihiko Hasegawa
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 data knotting and jointly construct polynomials instead of fixed-data VCA | Vanishing-polynomial classification · Iris, Wine, Vehicle, Vowel | accuracy: Iris 0.95 -> 0.96; Wine 0.98 -> 0.98; Vehicle 0.80 -> 0.80; Vowel 0.93 -> 0.92 | Mixed |
| Use data knotting and jointly construct polynomials instead of fixed-data VCA | Vanishing-polynomial classification · Iris, Wine, Vehicle, Vowel | test runtime (sec): Iris 6.6e-4 -> 3.6e-4; Wine 2.1e-3 -> 7.6e-4; Vehicle 6.7e-2 -> 1.6e-3; Vowel 2.3e-3 -> 1.5e-3 | Improved |
| Use data knotting and jointly construct polynomials instead of fixed-data VCA | Vanishing-polynomial feature extraction · Iris, Wine, Vehicle, Vowel | #features: Iris 62.8 -> 14.9; Wine 592.9 -> 100.5; Vehicle 4147.9 -> 121.5; Vowel 267.6 -> 191.0 | Improved |
| Use data knotting and jointly construct polynomials instead of fixed-data VCA | Vanishing-polynomial feature extraction · Iris, Wine, Vehicle, Vowel | #degree: Iris 2.0 -> 1.5; Wine 2.3 -> 1.7; Vehicle 2.5 -> 1.6; Vowel 1.8 -> 1.5 | Improved |
| Restrict feature extraction to half of the higher-degree polynomials (ablation) | Proposed · Iris, Wine, Vowel | accuracy: Iris 0.96 -> 0.75; Wine 0.98 -> 0.94; Vowel 0.92 -> 0.80 | Hurt |
| Restrict feature extraction to half of the higher-degree polynomials (ablation) | Proposed · Vehicle | accuracy: 0.80 -> 0.53 | Hurt |
| Restrict feature extraction to half of the higher-degree polynomials (ablation) | VCA · Iris, Wine, Vowel | accuracy: Iris 0.95 -> 0.59; Wine 0.98 -> 0.67; Vowel 0.93 -> 0.35 | Hurt |
| Restrict feature extraction to half of the higher-degree polynomials (ablation) | VCA · Vehicle | accuracy: 0.80 -> 0.60 | Improved |
| Represent the training data with discovered data knots rather than original points | 1-nearest-neighbor classification · Iris, Wine | accuracy: Iris 0.94 -> 0.95; Wine 0.95 -> 0.95 | No clear effect |
| Represent the training data with discovered data knots rather than original points | 1-nearest-neighbor classification · Vehicle, Vowel | accuracy: Vehicle 0.68 -> 0.60; Vowel 0.96 -> 0.76 | Hurt |
Closest papers by meaning
- Learning to compute Gröbner bases (NeurIPS 2024 · similarity 0.82)
- Approximating Latent Manifolds in Neural Networks via Vanishing Ideals (ICML 2025 · similarity 0.81)
- Machine Learning meets Algebraic Combinatorics: A Suite of Datasets Capturing Research-level Conjecturing Ability in Pure Mathematics (ICML 2025 · similarity 0.80)
- Fast exact recovery of noisy matrix from few entries: the infinity norm approach (NeurIPS 2025 · similarity 0.80)
- Exponent Relaxation of Polynomial Zonotopes and Its Applications in Formal Neural Network Verification (AAAI 2024 · similarity 0.79)
- Unsupervised Discovery of Formulas for Mathematical Constants (NeurIPS 2024 · similarity 0.79)
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.
