Research map / Learning theory and algorithms / Kernels and regression theory
The SKIM-FA Kernel: High-Dimensional Variable Selection and Nonlinear Interaction Discovery in Linear Time
Raj Agrawal, Tamara Broderick
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 SKIM-FA instead of Pairs Lasso | interaction variable selection · Synthetic p=1000, Equal setting | # wrong covariates selected: 213 -> 0 | Improved |
| Use SKIM-FA instead of MARS | interaction variable selection · Synthetic p=1000, Weak Main setting | # wrong covariates selected: 75 -> 9 | Improved |
| Use SKIM-FA instead of SPAM-2Stage | interaction variable selection · Synthetic p=1000, Weak Main setting | # correct covariates selected: 1 -> 5 | Improved |
| Use SKIM-FA instead of SPAM-2Stage when main effects are the only signal | estimation · Synthetic p=1000, Main Only setting | total SSE / signal variance: 0.17 -> 0.55 | Hurt |
| Use MARS-EMP (re-orthogonalized to empirical measure) instead of MARS-VANILLA (ablation) | MARS · Synthetic p=1000, Equal setting | total SSE / signal variance: 24.64 -> 0.43 | Improved |
| Use SKIM-FA instead of MARS-EMP | estimation · Synthetic p=1000, Weak Main setting | total SSE / signal variance: 0.78 -> 0.17 | Improved |
| Use SKIM-FA instead of MARS | variable selection · Bike Sharing + 1000 noise covariates | # wrong covariates selected: 119 -> 0 | Improved |
| Use SKIM-FA instead of MARS-Vanilla | estimation · Bike Sharing + 1000 noise covariates | total SSE: 10.209 -> 0.263 | Improved |
| Use SKIM-FA instead of SPAM-2Stage on real data | estimation · Bike Sharing + 1000 noise covariates | total SSE: 0.269 -> 0.263 | No clear effect |
| Use SKIM-FA instead of HierLasso on Main Effects Only | variable selection · Synthetic p=1000, Main-Only setting | # correct covariates selected: 4 -> 3 | Hurt |
Closest papers by meaning
- The Kernel Interaction Trick: Fast Bayesian Discovery of Pairwise Interactions in High Dimensions (ICML 2019 · similarity 0.91)
- kernelPSI: a Post-Selection Inference Framework for Nonlinear Variable Selection (ICML 2019 · similarity 0.89)
- Modelling Interactions in High-dimensional Data with Backtracking (JMLR 2016 · similarity 0.88)
- Permutation-Free High-Order Interaction Tests (ICML 2025 · similarity 0.86)
- Optimal Kernel Choice for Score Function-based Causal Discovery (ICML 2024 · similarity 0.84)
- Ab Initio Nonparametric Variable Selection for Scalable Symbolic Regression with Large $p$ (ICML 2025 · 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.
