Research map / Label-efficient and robust learning / Semi-supervised learning
Semi-Supervised Learning for FGVC With Out-of-Category Data
Ruoyi Du, Dongliang Chang, Zhanyu Ma, Kongming Liang, Yi-Zhe Song, Jun Hai Guo
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 |
|---|---|---|---|
| add triplet consistency regularization to relation-based pseudo-labeling (ablation) | RelMatch (from scratch) · Semi-Aves (U_in+U_out) | top-1 accuracy (%): 28.1 -> 31.2 | Improved |
| add label transfer via triplet consistency (ablation) | RelMatch (from scratch) · Semi-Aves (U_in+U_out) | top-1 accuracy (%): 31.2 -> 36.2 | Improved |
| add label transfer (ablation) | RelMatch (from scratch) · Semi-Fungi (U_in+U_out) | top-1 accuracy (%): 36.2 -> 39.6 | Improved |
| use deeper phylogenetic tree (depth 2 to 5) (ablation) | RelMatch (from scratch) · Semi-Aves (U_in+U_out) | top-1 accuracy (%): 25.4 -> 36.2 | Improved |
| use full tree depth 7 instead of depth 6 (ablation) | RelMatch (from scratch) · Semi-Fungi (U_in+U_out) | top-1 accuracy (%): 39.3 -> 39.6 | Improved |
| add out-of-distribution unlabeled data to FixMatch | FixMatch (from ImageNet) · Semi-Aves | top-1 accuracy (%): 57.4 -> 49.7 | Hurt |
| add out-of-distribution unlabeled data to RelMatch | RelMatch (from scratch) · Semi-Aves | top-1 accuracy (%): 34.5 -> 36.2 | Improved |
| add out-of-distribution unlabeled data to Pseudo-Label | Pseudo-Label (from scratch) · Semi-Aves | top-1 accuracy (%): 16.7 -> 12.2 | Hurt |
| add RelMatch to MoCo | MoCo (from scratch) · Semi-Aves (U_in+U_out) | top-1 accuracy (%): 38.9 -> 41.2 | Improved |
| add RelMatch to FixMatch | FixMatch (from ImageNet) · Semi-Aves (U_in) | top-1 accuracy (%): 57.4 -> 59.4 | Improved |
| add RelMatch to FixMatch with out-of-distribution data | FixMatch (from scratch) · Semi-Aves (U_in+U_out) | top-1 accuracy (%): 19.2 -> 32.3 | Improved |
| predict pair relations between labeled and unlabeled samples instead of U versus U (ablation) | relation pseudo-labeling · Semi-Aves out-of-distribution (relation accuracy) | relation accuracy (%): 70.1 -> 76.7 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- FixMatch — FixMatch: Simplifying Semi-Supervised Learning with Consistency and Confidence (NeurIPS 2020) · better on 6 of 6 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Large Scale Fine-Grained Categorization and Domain-Specific Transfer Learning (CVPR 2018 · similarity 0.87)
- Novel Class Discovery for Ultra-Fine-Grained Visual Categorization (CVPR 2024 · similarity 0.85)
- Democratizing Fine-grained Visual Recognition with Large Language Models (ICLR 2024 · similarity 0.84)
- Advancing Fine-Grained Classification by Structure and Subject Preserving Augmentation (NeurIPS 2024 · similarity 0.84)
- Scaling Up Semi-supervised Learning with Unconstrained Unlabelled Data (AAAI 2024 · similarity 0.83)
- Fine-grained Classes and How to Find Them (ICML 2024 · similarity 0.83)
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.
