Research map / Label-efficient and robust learning / Recognition and long-tail
Environment Diversification with Multi-head Neural Network for Invariant Learning
Bo-Wei Huang, Keng-Te Liao, Chang-Sheng Kao, Shou-De Lin
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 |
|---|---|---|---|
| set beta=0 (remove L_LI) in environment inference (ablation) | EDNIL · Adult-Confounded OOD | accuracy (%): 77.5 -> 41.3 | Hurt |
| set gamma=0 (remove L_IP feedback from invariant learner) (ablation) | EDNIL · Adult-Confounded OOD | accuracy (%): 77.5 -> 72.5 | Hurt |
| remove L_LI (beta=0) (ablation) | EDNIL · Synthetic regression OOD | mean square error: 2.253 -> 2.933 | Hurt |
| remove L_LI (beta=0) (ablation) | EDNIL · Synthetic regression stability | mean error: 2.218 -> 2.879 | Hurt |
| replace ERM with EDNIL on worst group | ERM · Waterbirds (Water,Land) | accuracy (%): 72.8 -> 78.6 | Improved |
| replace EIIL with EDNIL | EIIL · Waterbirds (Water,Land) | accuracy (%): 68.6 -> 78.6 | Improved |
| replace ERM with EDNIL | ERM · CMNIST color noise 0.9 | accuracy (%): 21.7 -> 75.2 | Improved |
| replace ERM with EDNIL | ERM · Adult-Confounded OOD | accuracy (%): 40.7 -> 77.5 | Improved |
| replace ERM with EDNIL on bias-misaligned subset | ERM · SNLI misaligned | accuracy (%): 52.6 -> 54.5 | Improved |
| replace EIIL with EDNIL | EIIL · Adult-Confounded OOD | accuracy (%): 70.2 -> 77.5 | Improved |
| replace EIIL with EDNIL | EIIL · CMNIST color noise 0.9 | accuracy (%): 63.1 -> 75.2 | Improved |
| replace KerHRM with EDNIL | KerHRM · Adult-Confounded OOD | accuracy (%): 67.9 -> 77.5 | Improved |
Closest papers by meaning
- Environment Inference for Invariant Learning (ICML 2021 · similarity 0.87)
- Provably Invariant Learning without Domain Information (ICML 2023 · similarity 0.86)
- ZIN: When and How to Learn Invariance Without Environment Partition? (NeurIPS 2022 · similarity 0.85)
- Sufficient Invariant Learning for Distribution Shift (CVPR 2025 · similarity 0.85)
- Diagnosing and Rectifying Fake OOD Invariance: A Restructured Causal Approach (AAAI 2024 · similarity 0.83)
- The Implicit Bias of Heterogeneity towards Invariance: A Study of Multi-Environment Matrix Sensing (NeurIPS 2024 · similarity 0.82)
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.
