Research map / Label-efficient and robust learning / Recognition and long-tail
Variational Connectionist Temporal Classification
Linlin Chao, Jingdong Chen, Wei Chu
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 |
|---|---|---|---|
| replace vanilla CTC with Var-CTC | CRNN-CTC · IC03 | recall at 98% precision (%): 58.0 -> 90.3 | Improved |
| replace vanilla CTC with Var-CTC | CRNN-CTC · IC13 | recall at 98% precision (%): 78.5 -> 92.1 | Improved |
| replace vanilla CTC with Var-CTC | CRNN-CTC · IIIT5K | recall at 98% precision (%): 30.2 -> 66.4 | Improved |
| replace vanilla CTC with Var-CTC | CRNN-CTC · SVT | recall at 98% precision (%): 34.3 -> 85.1 | Improved |
| add approximate posterior of blank on top of Mml-CTC (Var-CTC vs Mml-CTC) (ablation) | Mml-CTC · SVT | recall at 98% precision (%): 57.7 -> 85.1 | Improved |
| add approximate posterior of blank on top of Mml-CTC (Var-CTC vs Mml-CTC) (ablation) | Mml-CTC · IIIT5K | recall at 98% precision (%): 68.7 -> 66.4 | No clear effect |
| replace vanilla CTC with Var-CTC | CRNN-CTC (ours) · IIIT5K | sequence accuracy (%): 79.4 -> 78.7 | No clear effect |
| replace vanilla CTC with Var-CTC | CRNN-CTC (ours) · IC13 | sequence accuracy (%): 89.6 -> 88.7 | Hurt |
| replace vanilla CTC with Mml-CTC | CRNN-CTC (ours) · IIIT5K | sequence accuracy (%): 79.4 -> 81.3 | Improved |
| replace vanilla CTC with Var-CTC | ResNet-CTC · IAM handwriting | word error rate (%): 23.8 -> 24.0 | No clear effect |
| replace vanilla CTC with Var-CTC | CTC · Seven scene text datasets (ASTER setup) | recall at 98% precision (%): 67.57 -> 81.92 | Improved |
| replace vanilla CTC with Var-CTC | CRNN-CTC · IIIT5K | Replace error share (%): 49% -> 40% | Improved |
Closest papers by meaning
- Align With Purpose: Optimize Desired Properties in CTC Models with a General Plug-and-Play Framework (ICLR 2024 · similarity 0.87)
- Self-Distillation Regularized Connectionist Temporal Classification Loss for Text Recognition: A Simple Yet Effective Approach (AAAI 2024 · similarity 0.87)
- CR-CTC: Consistency regularization on CTC for improved speech recognition (ICLR 2025 · similarity 0.87)
- TSVC: Tripartite Learning with Semantic Variation Consistency for Robust Image-Text Retrieval (AAAI 2025 · similarity 0.81)
- Variational Supervised Contrastive Learning (NeurIPS 2025 · similarity 0.80)
- MLC-NC: Long-Tailed Multi-Label Image Classification Through the Lens of Neural Collapse (AAAI 2025 · similarity 0.80)
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.
