Research map / Learning theory and algorithms / Matrix and tensor methods
Generalized Sparse Learning of Linear Models Over the Complete Subgraph Feature Set
Ichigaku Takigawa, Hiroshi Mamitsuka
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 L1-LogReg direct optimization instead of Adaboost/LPBoost boosting | subgraph learner · RAND | convergence iterations: 20 iterations vs around 100 and 50 | Improved |
| use hinge loss (LPBoost) instead of logistic loss | L1-LogReg · RAND | final error: 0.1704 vs 0.1758 | Mixed |
| use L1-LogReg instead of boosting with same feature count | subgraph selection · RAND | subgraph size distribution | Improved |
| use L1-LogReg instead of Adaboost | subgraph learner · CPDB (10-fold CV) | number of features: 180.3 -> 80.3 | Improved |
| use inexact pruning (skip repeated characteristic vectors, size > 3) instead of exact (ablation) | L1-LogReg · CPDB (10-fold CV) | test accuracy: 0.774 -> 0.783 | Improved |
| use inexact pruning instead of exact (ablation) | L1-LogReg · CPDB (10-fold CV) | CPU time (sec): 2078.88 -> 695.39 | Improved |
| add L2 penalty (elastic net) with increasing lambda2 (ablation) | L1-LogReg · CPDB (10-fold CV) | test accuracy: 0.759 -> 0.755 -> 0.745 | Hurt |
| increase lambda1 (ablation) | L1-LogReg exact · CPDB (10-fold CV) | test accuracy: 0.755 -> 0.774 -> 0.762 -> 0.750 -> 0.733 | Mixed |
| increase lambda1 (ablation) | L1-LogReg exact · CPDB (10-fold CV) | CPU time (sec): 8918.94 -> 97.29 | Improved |
| use fingerprint glmnet instead of subgraph L1-LogReg | L1-LogReg · CPDB (10-fold CV) | test accuracy: 0.774 -> 0.739 | Hurt |
Closest papers by meaning
- Joint Edge-Model Sparse Learning is Provably Efficient for Graph Neural Networks (ICLR 2023 · similarity 0.89)
- Sparse Logistic Regression Learns All Discrete Pairwise Graphical Models (NeurIPS 2019 · similarity 0.89)
- Sample-Efficient L0-L2 Constrained Structure Learning of Sparse Ising Models (AAAI 2021 · similarity 0.88)
- Learning on Hypergraphs with Sparsity (TPAMI 2020 · similarity 0.88)
- Quantifying Distributional Invariance in Causal Subgraph for IRM-Free Graph Generalization (NeurIPS 2025 · similarity 0.87)
- Graph Sparsification via Mixture of Graphs (ICLR 2025 · 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.
