Research map / Label-efficient and robust learning / Domain adaptation and generalization
Robust Learning with the Hilbert-Schmidt Independence Criterion
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| train with HSIC loss instead of standard loss | linear regression on Bike Sharing · Bike Sharing target seasons | residual variance | Mixed |
| train with HSIC loss instead of cross-entropy | DeepLoc · Cell Out of Sample Target1 | class balanced accuracy: 98.4 -> 99.2 | Helped |
| train with HSIC loss instead of cross-entropy | DeepLoc · Cell Out of Sample Target2 | class balanced accuracy: 98.1 -> 98.8 | Helped |
| train with HSIC loss instead of cross-entropy | DeepLoc · Cell Out of Sample Target3 | class balanced accuracy: 91.7 -> 93.4 | Helped |
| train with HSIC loss instead of cross-entropy | DeepLoc · Cell Out of Sample Target4 | class balanced accuracy: 93.8 -> 95.3 | Helped |
| add test-time augmentation (ablation) | HSIC-trained DeepLoc · Cell Out of Sample Target1 | class balanced accuracy: 99.2 -> 99.2 | No clear effect |
| add test-time augmentation (ablation) | HSIC-trained DeepLoc · Cell Out of Sample Target2 | class balanced accuracy: 98.8 -> 98.9 | Helped |
| add test-time augmentation (ablation) | HSIC-trained DeepLoc · Cell Out of Sample Target3 | class balanced accuracy: 93.4 -> 93.4 | No clear effect |
| add test-time augmentation (ablation) | HSIC-trained DeepLoc · Cell Out of Sample Target4 | class balanced accuracy: 95.3 -> 95.4 | Helped |
| increase mini-batch size from 32 to 64 or 128 (ablation) | HSIC loss on synthetic linear regression · synthetic regression | MSE | Hurt |
Closest papers by meaning
- On Statistical Learning Theory for Distributional Inputs (ICML 2024 · similarity 0.86)
- Label Noise: Ignorance Is Bliss (NeurIPS 2024 · similarity 0.86)
- Pseudo-Labeling for Unsupervised Domain Adaptation with Kernel GLMs (NeurIPS 2026 · similarity 0.86)
- Robustly Learning Single-Index Models via Alignment Sharpness (ICML 2024 · similarity 0.86)
- Robust Loss Functions for Training Decision Trees with Noisy Labels (AAAI 2024 · similarity 0.86)
- Out-Of-Domain Unlabeled Data Improves Generalization (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.