Research map / Label-efficient and robust learning / Domain adaptation and generalization
Invariance Learning based on Label Hierarchy
Shoji Toyota, Kenji Fukumizu
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 label-hierarchy invariance learning with higher-level data and Our CV I instead of ERM | Synthesized Data 1 · e*=50, test e=-e* | test accuracy: .051 (.101) -> .699 (.255) | Improved |
| Use label-hierarchy invariance learning with higher-level data and Our CV II instead of FT | Synthesized Data 1 · e*=50, test e=-e* | test accuracy: .000 (.000) -> .699 (.255) | Improved |
| Use CV II rather than CV I as domains approach the target domain | Synthesized Data 2 · ead=1, test e=-e* | test accuracy: .218 (.087) -> .502 (.149) | Improved |
| Use CV II rather than CV I when additional-data domains are distant from the target | Synthesized Data 2 · ead=-9, test e=-e* | test accuracy: .529 (.128) -> .527 (.152) | No clear effect |
| Use label-hierarchy invariance learning with CV II instead of ERM | Hierarchical Colored MNIST · e=0.9 | test accuracy: .177 (.006) -> .707 (.012) | Improved |
| Use CV II rather than CV I for hyperparameter selection | Hierarchical Colored MNIST · e=0.9 | test accuracy: .644 (.011) -> .707 (.012) | Improved |
| Use label-hierarchy invariance learning rather than ERM | Bird recognition · e2 | test accuracy: .317 (.044) -> .727 (.062) | Improved |
| Use the proposed method rather than FT | Bird recognition · e1 | test accuracy: .909 (.012) -> .897 (.020) | Mixed |
| Use CV II rather than CV I for hyperparameter selection | Bird recognition · e2 | test accuracy: .727 (.062) -> .727 (.062) | No clear effect |
| Use the proposed method with CV II rather than LOD-CV | Hierarchical Colored MNIST · e=0.9 | test accuracy: .334 (.029) -> .707 (.012) | Improved |
Closest papers by meaning
- Learning Invariances in Neural Networks from Training Data (NeurIPS 2020 · similarity 0.89)
- Exploiting Invariance in Training Deep Neural Networks (AAAI 2022 · similarity 0.88)
- Continuous Invariance Learning (ICLR 2024 · similarity 0.88)
- Unsupervised Representation Learning by Invariance Propagation (NeurIPS 2020 · similarity 0.87)
- Empowering Graph Invariance Learning with Deep Spurious Infomax (ICML 2024 · similarity 0.85)
- Learning Conditional Invariances through Non-Commutativity (ICLR 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.
