Research map / Optimization / Optimizers
Multi-output Polynomial Networks and Factorization Machines
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 group-sparse l1/l2 or l1/l_infinity penalties instead of l1 (ablation) | Multi-output PN with output-layer refitting · vowel and letter | test accuracy: 87.83 -> 89.57 | Helped |
| Use group-sparse penalties instead of l1 (ablation) | Multi-output PN · segment and satimage | test accuracy | No clear effect |
| Refit hidden and output layers instead of only the output layer (ablation) | Multi-output PN · vowel | test accuracy: 85.22 -> 89.57 | Helped |
| Use MSH instead of ML loss (ablation) | Multi-output PN, full refitting · vowel | test accuracy: 89.57 -> 86.96 | Hurt |
| Use MSH instead of ML loss (ablation) | Multi-output PN, full refitting · segment | test accuracy: 96.71 -> 96.01 | Mixed |
| Initialize basis refinement with the l1 solution instead of random initialization (ablation) | l1/l_infinity basis-vector selection · basis-selection experiments | multiplicative approximation factor | Helped |
| Refine the l1-initialized basis vector iteratively (ablation) | l1/l_infinity basis-vector selection · basis-selection experiments | multiplicative approximation factor | Helped |
| Train multi-output FMs with ordinal McRank instead of single-output FMs | Factorization machines · MovieLens 100k | nDCG@1: 0.71 -> 0.75 | Helped |
| Train multi-output FMs with ordinal McRank instead of single-output FMs | Factorization machines · MovieLens 1M | nDCG@1: 0.75 -> 0.76 | Helped |
| Use multi-output PNs instead of one-vs-rest PNs | Polynomial networks · multi-class datasets | test accuracy | Helped |
| Use full hidden-layer refitting instead of output-layer refitting (ablation) | Multi-output PN · multi-class datasets | test accuracy | Mixed |
Closest papers by meaning
- Multiclass Loss Geometry Matters for Generalization of Gradient Descent in Separable Classification (NeurIPS 2025 · similarity 0.85)
- TMPNN: High-Order Polynomial Regression Based on Taylor Map Factorization (AAAI 2024 · similarity 0.85)
- Learning Hierarchical Polynomials with Three-Layer Neural Networks (ICLR 2024 · similarity 0.85)
- Neural network learns low-dimensional polynomials with SGD near the information-theoretic limit (NeurIPS 2024 · similarity 0.85)
- Generalized Tensor Decomposition for Understanding Multi-Output Regression under Combinatorial Shifts (NeurIPS 2024 · similarity 0.85)
- How Universal Polynomial Bases Enhance Spectral Graph Neural Networks: Heterophily, Over-smoothing, and Over-squashing (ICML 2024 · similarity 0.85)
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.