Research map / Learning theory and algorithms / Matrix and tensor methods
High-Dimensional Interactions Detection with Sparse Principal Hessian Matrix
Cheng Yong Tang, Ethan X. Fang, Yuexiao Dong
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 one-step principal Hessian (ADMM) instead of two-step screening IP | interaction detection · Simulation Model 8, d=100 | true positive rate: 75.5% -> 97.0% | Improved |
| use ADMM instead of IP as noise sigma rises | interaction detection · Simulation Model 8 (0,1.25), d=100 | true positive rate: 48.5% -> 95.0% | Improved |
| use ADMM instead of RAMP when no hierarchy exists | interaction detection · Simulation Model 2, d=100 | true positive rate: 1.00% -> 99.0% | Improved |
| use RAMP hierarchy instead of ADMM when strong main effects exist | interaction detection · Simulation Model 6, d=100 | true positive rate: 99.5% vs 98.5% | Hurt |
| use ADMM instead of RAMP with weak main effects | interaction detection · Simulation Model 7, d=100 | true positive rate: 1.00% -> 100.0% | Improved |
| raise dimension d from 100 to 300 for ADMM | ADMM · Simulation Model 2 (0,0.1) | true positive rate: 99.0% -> 92.5% | Hurt |
| use ADMM instead of IP/RAMP on breast tumor data | interaction detection · GPL96 breast tumor | classification error: 8.15% -> 6.13% | Improved |
| use ADMM instead of RAMP on Ewing Sarcoma | interaction detection · GPL96 Ewing Sarcoma | classification error: 19.72% -> 5.58% | Improved |
| use ADMM instead of IP on Ewing Sarcoma | interaction detection · GPL96 Ewing Sarcoma | classification error: 8.90% -> 5.58% | Improved |
| use ADMM instead of IP for running time | interaction detection · Simulation Model 1, d=300 | running time: 0.34 s -> 2.13 s | Hurt |
Closest papers by meaning
- The SKIM-FA Kernel: High-Dimensional Variable Selection and Nonlinear Interaction Discovery in Linear Time (JMLR 2023 · similarity 0.88)
- Modelling Interactions in High-dimensional Data with Backtracking (JMLR 2016 · similarity 0.87)
- The Kernel Interaction Trick: Fast Bayesian Discovery of Pairwise Interactions in High Dimensions (ICML 2019 · similarity 0.87)
- WHInter: A Working set algorithm for High-dimensional sparse second order Interaction models (ICML 2018 · similarity 0.86)
- Fast and More Powerful Selective Inference for Sparse High-Order Interaction Model (AAAI 2022 · similarity 0.86)
- Multi-way Interacting Regression via Factorization Machines (NeurIPS 2017 · 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.
