Research map / Learning theory and algorithms / Prediction and decision losses
Beyond Trees: Classification with Sparse Pairwise Dependencies
Yaniv Tenzer, Amit Moscovich, Mary Frances Dorn, Boaz Nadler, Clifford Spiegelman
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 |
|---|---|---|---|
| add HSIC-based filtering of near-independent bivariate densities (ablation) | SLB · General BN, Gaussian, unbalanced, n=1000 | misclassification error (%): 5.29 -> 4.11 | Improved |
| add HSIC-based filtering of near-independent bivariate densities (ablation) | SLB · General BN, Gaussian, balanced, n=200 | misclassification error (%): 12.3 -> 11.7 | Improved |
| add HSIC-based filtering of near-independent bivariate densities (ablation) | SLB · General BN, Gaussian, balanced, n=1000 | misclassification error (%): 3.73 -> 3.67 | No clear effect |
| add HSIC-based filtering of near-independent bivariate densities (ablation) | SLB · General BN, Gaussian, unbalanced, n=200 | misclassification error (%): 15.1 -> 15.1 | No clear effect |
| add bivariate log-density features to univariate ones | SLB vs LU · Ionosphere | balanced error rate (%): 45.4 -> 7.5 | Improved |
| add bivariate log-density features to univariate ones | SLB vs LU · Pima | balanced error rate (%): 49.2 -> 28.6 | Improved |
| replace tree-structured TAN with SLB under general BNs | SLB · General BN, Gaussian, balanced, n=1000 | misclassification error (%): 10.8 -> 3.67 | Improved |
| replace TAN with SLB on forest-structured Gaussian data | SLB · Forest BN, Gaussian, balanced, n=200 | misclassification error (%): 8.17 -> 15.4 | Hurt |
| increase training sample size from 200 to 1000 | SLB · Forest BN, Gaussian, balanced | misclassification error (%): 15.4 -> 5.11 | Improved |
| replace SLB with random forest on complex forest data | SLB · Forest BN, complex, balanced, n=1000 | misclassification error (%): 27.6 -> 23.8 | Hurt |
| replace TAN with SLB on real data | SLB · Ionosphere | balanced error rate (%): 7.6 -> 7.5 | No clear effect |
| replace NB with SLB on Ozone | SLB · Ozone | balanced error rate (%): 24.9 -> 32.3 | Hurt |
Closest papers by meaning
- Hierarchical Shrinkage: Improving the accuracy and interpretability of tree-based models. (ICML 2022 · similarity 0.87)
- Sparse Logistic Regression Learns All Discrete Pairwise Graphical Models (NeurIPS 2019 · similarity 0.87)
- Random Composite Forests (AAAI 2016 · similarity 0.87)
- Optimal Classification Trees for Continuous Feature Data Using Dynamic Programming with Branch-and-Bound (AAAI 2025 · similarity 0.86)
- Learning Small Decision Trees with Few Outliers: A Parameterized Perspective (AAAI 2024 · similarity 0.86)
- Symbolic Regression Enhanced Decision Trees for Classification Tasks (AAAI 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.
