Research map / Label-efficient and robust learning / Recognition and long-tail
Regularized Discriminative Alignment for Deep Representations under Label Shift
Hengchao Shi, Boen Jiang, Guanhua Fang, Wen Yu, Ming Zheng
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 LDA instead of PCA for summary functions (ablation) | RDALS · MNIST | weight estimation MSE: 2.938 -> 0.018 | Improved |
| use LDA instead of random projection for summary functions (ablation) | RDALS · MNIST | weight estimation MSE: 2.195 -> 0.018 | Improved |
| use LDA instead of PCA for summary functions (ablation) | RDALS · CIFAR-10 | weight estimation MSE: 0.573 -> 0.051 | Improved |
| use LDA instead of random projection for summary functions (ablation) | RDALS · CIFAR-10 | weight estimation MSE: 0.736 -> 0.051 | Improved |
| use LDA instead of PCA for summary functions (ablation) | RDALS · CIFAR-100 | weight estimation MSE: 3.248 -> 1.686 | Improved |
| use LDA instead of random projection for summary functions (ablation) | RDALS · CIFAR-100 | weight estimation MSE: 4.063 -> 1.686 | Improved |
| use regularized SOCP solver instead of direct inverse (LDA) (ablation) | RDALS · CIFAR-10 | weight estimation MSE: 4.785 -> 0.056 | Improved |
| use regularized SOCP solver instead of direct inverse (LDA) (ablation) | RDALS · MNIST | weight estimation MSE: 3.879 -> 0.014 | Improved |
| use regularized SOCP solver instead of direct inverse (LDA) (ablation) | RDALS · CIFAR-100 | weight estimation MSE: 29.325 -> 1.703 | Improved |
| use regularized SOCP solver instead of direct inverse (PCA) (ablation) | RDALS with PCA · CIFAR-10 | weight estimation MSE: 7.664 -> 0.489 | Improved |
| vary regularization lambda from 1e-6 to 1e-1 (ablation) | RDALS · CIFAR-10 (N=10000) | weight estimation MSE: 4.555e-2 -> 4.857e-2 | No clear effect |
| RDALS versus CPMCN running time | RDALS vs CPMCN · CIFAR-10 (N=10000) | running time (s): 167.82 -> 74.18 | Improved |
Closest papers by meaning
- Regularized Learning for Domain Adaptation under Label Shifts (ICLR 2019 · similarity 0.89)
- RLSbench: Domain Adaptation Under Relaxed Label Shift (ICML 2023 · similarity 0.87)
- LTF: A Label Transformation Framework for Correcting Label Shift (ICML 2020 · similarity 0.87)
- CASUAL: Conditional Support Alignment for Domain Adaptation with Label Shift (AAAI 2025 · similarity 0.85)
- Stabilizing Sample Similarity in Representation via Mitigating Random Consistency (ICML 2025 · similarity 0.85)
- Addressing Label Shift in Distributed Learning via Entropy Regularization (ICLR 2025 · 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.
