Research map / Probabilistic and scientific modeling / Causal discovery
Long-Tailed Classification by Keeping the Good and Removing the Bad Momentum Causal Effect
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 |
|---|---|---|---|
| add TDE inference to de-confounded model (ablation) | De-confound · ImageNet-LT | overall top-1 accuracy: 48.6 -> 51.8 | Helped |
| add TDE inference to de-confounded model (ablation) | De-confound · ImageNet-LT | few-shot top-1 accuracy: 14.7 -> 31.6 | Helped |
| add TDE inference to de-confounded model (ablation) | De-confound · ImageNet-LT | many-shot top-1 accuracy: 67.9 -> 62.7 | Hurt |
| add TDE inference to cosine classifier (ablation) | Cosine · ImageNet-LT | overall top-1 accuracy: 47.6 -> 50.5 | Helped |
| add TDE inference to capsule classifier (ablation) | Capsule · ImageNet-LT | overall top-1 accuracy: 46.5 -> 50.6 | Helped |
| use de-confounded normalized multi-head classifier instead of linear classifier (ablation) | Baseline · ImageNet-LT | overall top-1 accuracy: 45.0 -> 48.6 | Helped |
| add de-confounded training and TDE (ablation) | Baseline · LT-CIFAR-100 imbalance 100 | top-1 accuracy: 42.6 (BBN) -> 44.1 | Helped |
| add TDE inference to de-confounded model (ablation) | De-confound · LT-CIFAR-10 imbalance 100 | top-1 accuracy: 71.7 -> 80.6 | Helped |
| add TDE inference to de-confounded model (ablation) | De-confound · LT-CIFAR-10 imbalance 10 | top-1 accuracy: 86.8 -> 88.5 | Helped |
| add TDE inference to de-confounded model (ablation) | De-confound · LVIS v0.5 | mask AP: 25.7 -> 28.4 | Helped |
| add TDE inference to de-confounded model (ablation) | De-confound · LVIS v1.0 | mask AP: 23.5 -> 27.1 | Helped |
| increase TDE parameter alpha (ablation) | De-confound-TDE · ImageNet-LT val | accuracy by split | Mixed |
Closest papers by meaning
- Long-Tailed Learning as Multi-Objective Optimization (AAAI 2024 · similarity 0.84)
- Decoupled Optimisation for Long-Tailed Visual Recognition (AAAI 2024 · similarity 0.83)
- Rethinking the Bias of Foundation Model under Long-tailed Distribution (ICML 2025 · similarity 0.83)
- Feature Fusion from Head to Tail for Long-Tailed Visual Recognition (AAAI 2024 · similarity 0.83)
- Learning from Neighbors: Category Extrapolation for Long-Tail Learning (CVPR 2025 · similarity 0.83)
- Exploring Weight Balancing on Long-Tailed Recognition Problem (ICLR 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.