Research map / Probabilistic and scientific modeling / Variational and generative sampling
Revisiting Probability Distribution Assumptions for Information Theoretic Feature Selection
Yuan Sun, Wei Wang, Michael Kirley, Xiaodong Li, Jeffrey Chan
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 |
|---|---|---|---|
| replace FID with GMD assumption (ablation) | MI method J1,0 · 29 UCI, SVM | average rank: 2.52 -> 1.79 | Improved |
| replace GMD with AMD assumption (ablation) | MI method J1,0 · 29 UCI, SVM | average rank: 1.79 -> 1.31 | Improved |
| replace GMD with AMD assumption (ablation) | MI method J1,1 · 29 UCI, KNN | average rank: 1.69 -> 1.24 | Improved |
| replace FID with AMD assumption (ablation) | VI method J1 · 29 UCI, KNN | average rank: 2.28 -> 1.38 | Improved |
| replace GMD with AMD assumption (ablation) | VI method J1 · 29 UCI, KNN | average rank: 1.93 -> 1.38 | Improved |
| increase order k from 0 to 1 (J0,0 to J1,0) with GMD (ablation) | GMD-based MI · 29 UCI, SVM | average rank: 3.34 -> 1.59 | Improved |
| increase order from J1,1 to J2,1 with GMD (ablation) | GMD-based MI · 29 UCI, KNN | average rank: 2.20 -> 1.55 | Improved |
| increase order from J1,1 to J2,1 with AMD (ablation) | AMD-based MI · 29 UCI, SVM | average rank: 2.38 -> 1.79 | Improved |
| increase order from J1,0 to J1,1 with AMD (ablation) | AMD-based MI · 29 UCI, KNN | average rank: 2.14 -> 2.20 | Hurt |
| increase order k (ablation) | GMD/AMD-based MI · Computation cost | time | Hurt |
| AMD J2,1 versus MRI | feature selection · 29 UCI, SVM | average rank: 2.86 -> 2.34 | Improved |
| AMD J2,1 versus KNN-evaluated AMD VI J1 | feature selection · 29 UCI, KNN | average rank: 2.41 -> 2.65 | Mixed |
Closest papers by meaning
- Variational Information Maximization for Feature Selection (NeurIPS 2016 · similarity 0.92)
- A Rigorous Information-Theoretic Definition of Redundancy and Relevancy in Feature Selection Based on (Partial) Information Decomposition (JMLR 2023 · similarity 0.88)
- Flow-based Variational Mutual Information: Fast and Flexible Approximations (ICLR 2025 · similarity 0.87)
- Estimating Conditional Mutual Information for Dynamic Feature Selection (ICLR 2024 · similarity 0.85)
- Mutual Information Estimation via $f$-Divergence and Data Derangements (NeurIPS 2024 · similarity 0.85)
- InfoNet: Neural Estimation of Mutual Information without Test-Time Optimization (ICML 2024 · similarity 0.84)
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.
