Research map / Label-efficient and robust learning / Recognition and long-tail
Is Fairness Only Metric Deep? Evaluating and Addressing Subgroup Gaps in Deep Metric Learning
Natalie Dullerud, Karsten Roth, Kimia Hamidieh, Nicolas Papernot, Marzyeh Ghassemi
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 |
|---|---|---|---|
| Train the upstream DML embedding with manually imbalanced rather than balanced classes | Margin loss + distance-weighted sampling · CUB200-2011 (bird color) | upstream Recall@1 gap: 0.017\pm 0.007 -> 0.212\pm 0.029 | Hurt |
| Train the upstream DML embedding with manually imbalanced rather than balanced classes | Margin loss + distance-weighted sampling, followed by a balanced downstream classifier · CUB200-2011 (bird color) | downstream accuracy gap: 0.014\pm 0.002 -> 0.131\pm 0.027 | Hurt |
| Add sensitive-attribute embedding and partial de-correlation | Margin loss + distance-weighted sampling · CelebA (skintone) | downstream accuracy gap: 0.082 ± 0.005 -> 0.131 ± 0.006 | Hurt |
| Add sensitive-attribute embedding and partial de-correlation | Margin loss + distance-weighted sampling · LFW (race) | downstream accuracy gap: 0.012 ± 0.009 -> 0.009± 0.003 | Improved |
| Add sensitive-attribute embedding and partial de-correlation | Margin loss + distance-weighted sampling · CUB200-2011 (bird color; manually imbalanced) | downstream precision gap: 0.333 ± 0.049 -> 0.317 ± 0.046 | Improved |
| Add sensitive-attribute embedding and partial de-correlation | Triplet loss + distance-weighted sampling · CUB200-2011 (bird color; manually imbalanced) | downstream accuracy gap: 0.154 ± 0.029 -> 0.148 ± 0.049 | Improved |
Closest papers by meaning
- Revisiting Training Strategies and Generalization Performance in Deep Metric Learning (ICML 2020 · similarity 0.90)
- Improving Deep Metric Learning by Divide and Conquer (TPAMI 2021 · similarity 0.88)
- DiVA: Diverse Visual Feature Aggregation for Deep Metric Learning (ECCV 2020 · similarity 0.88)
- Deep Disentangled Metric Learning (AAAI 2025 · similarity 0.87)
- Threshold-Consistent Margin Loss for Open-World Deep Metric Learning (ICLR 2024 · similarity 0.86)
- The Fragility of Fairness: Causal Sensitivity Analysis for Fair Machine Learning (NeurIPS 2024 · similarity 0.86)
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.
