Research map / Label-efficient and robust learning / Recognition and long-tail
Understanding Medical Conversations with Scattered Keyword Attention and Weak Supervision from Responses
Xiaoming Shi, Haifeng Hu, Wanxiang Che, Zhongqian Sun, Ting Liu, Junzhou Huang
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 label-embedding attention (ablation) | TextCNN · Medical slot filling test | Micro F1: 70.69 -> 73.07 | Improved |
| add label-embedding attention (ablation) | Star-Transformer · Medical slot filling test | Micro F1: 74.04 -> 76.90 | Improved |
| add label-embedding attention (ablation) | DRNN · Medical slot filling test | Micro F1: 74.83 -> 76.07 | Improved |
| add label-embedding attention (ablation) | TextCNN · Medical slot filling test | Macro F1: 56.95 -> 62.40 | Improved |
| add weak supervision pre-training on top of attention (ablation) | TextCNN+A · Medical slot filling test | Micro F1: 73.07 -> 78.76 | Improved |
| add weak supervision pre-training on top of attention (ablation) | DRNN+A · Medical slot filling test | Micro F1: 76.07 -> 81.15 | Improved |
| add weak supervision pre-training on top of attention (ablation) | Star-Transformer+A · Medical slot filling test | Turn accuracy: 53.20 -> 52.80 | Hurt |
| add weak supervision pre-training on top of attention (ablation) | Star-Transformer+A · Medical slot filling test | Micro F1: 76.90 -> 78.05 | Improved |
| add weak supervision pre-training on top of BERT features (ablation) | TextCNN+B · Medical slot filling test | Micro F1: 71.22 -> 75.56 | Improved |
| add BERT fixed features (ablation) | TextCNN · Medical slot filling test | Micro F1: 70.69 -> 71.22 | Improved |
| use response concepts instead of query concepts as weak supervision (ablation) | TextCNN+WS · Medical slot filling test | Micro F1: 53.64 -> 72.59 | Improved |
| use response concepts instead of query concepts as weak supervision (ablation) | TextCNN+WS · Medical slot filling test | Turn accuracy: 27.50 -> 53.30 | Improved |
Closest papers by meaning
- DiN: Diffusion Model for Robust Medical VQA with Semantic Noisy Labels (CVPR 2025 · similarity 0.84)
- Uni-Med: A Unified Medical Generalist Foundation Model For Multi-Task Learning Via Connector-MoE (NeurIPS 2024 · similarity 0.83)
- SMMILE: An expert-driven benchmark for multimodal medical in-context learning (NeurIPS 2025 · similarity 0.82)
- Bringing CLIP to the Clinic: Dynamic Soft Labels and Negation-Aware Learning for Medical Analysis (CVPR 2025 · similarity 0.81)
- MIMO: A Medical Vision Language Model with Visual Referring Multimodal Input and Pixel Grounding Multimodal Output (CVPR 2025 · similarity 0.81)
- Alignment, Mining and Fusion: Representation Alignment with Hard Negative Mining and Selective Knowledge Fusion for Medical Visual Question Answering (CVPR 2025 · similarity 0.81)
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.
