Research map / Learning theory and algorithms / Optimal transport
Enhancing Interpretability in CLIP with Optimal Transport-based Submodular Optimization for Ophthalmic Imaging
Sihong Lu, Jian Yang, Lei Luo
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 visual and text transport costs to the OT stage (ablation) | OTSMDL · IDRiD | Insertion AUC: 0.406 -> 0.624 | Improved |
| add visual and text transport costs to the OT stage (ablation) | OTSMDL · OCTDL | Insertion AUC: 0.336 -> 0.530 | Improved |
| add text transport cost to visual-only cost (ablation) | OTSMDL · IDRiD | Insertion AUC: 0.487 -> 0.624 | Improved |
| add visual transport cost to text-only cost (ablation) | OTSMDL · IDRiD | Insertion AUC: 0.601 -> 0.624 | Improved |
| use unbalanced OT instead of balanced OT (ablation) | OTSMDL · IDRiD | Insertion AUC: 0.426 -> 0.624 | Improved |
| use unbalanced OT instead of balanced OT (ablation) | OTSMDL · IDRiD | Deletion AUC: 0.313 -> 0.226 | Improved |
| use unbalanced OT instead of balanced OT (ablation) | OTSMDL · OCTDL | Insertion AUC: 0.346 -> 0.530 | Improved |
| use score-guided source distribution instead of uniform a (ablation) | OTSMDL · IDRiD | Insertion AUC: 0.603 -> 0.624 | Improved |
| use score-guided source distribution instead of uniform a (ablation) | OTSMDL · OCTDL | Insertion AUC: 0.529 -> 0.530 | No clear effect |
| add importance, contrast and uniqueness scores to submodular initialization (ablation) | OTSMDL · IDRiD | Insertion AUC: 0.566 -> 0.624 | Improved |
| add importance, contrast and uniqueness scores to submodular initialization (ablation) | OTSMDL · OCTDL | Insertion AUC: 0.506 -> 0.530 | Improved |
| use uniqueness score alone instead of importance alone (ablation) | OTSMDL · IDRiD | Insertion AUC: 0.524 -> 0.612 | 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.
- RISE — RISE: Robust Individualized Decision Learning with Sensitive Variables (NeurIPS 2022) · better on 4 of 4 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
- FOCUS: Benchmarking Retinal Model Generalization from Foundation Vision Encoders to Multimodal LLMs (NeurIPS 2026 · similarity 0.86)
- BIOMEDICA: An Open Biomedical Image-Caption Archive, Dataset, and Vision-Language Models Derived from Scientific Literature (CVPR 2025 · similarity 0.85)
- A World Model of Radiologist Reading for Medical Image Representation Learning (NeurIPS 2026 · similarity 0.85)
- Rethinking Visual Attribution for Chest X-ray Reasoning in Large Vision Language Models (NeurIPS 2026 · similarity 0.85)
- CRAFT: Causal Responsibility and Failure Tracing in Medical Vision Language Models (NeurIPS 2026 · similarity 0.84)
- Incomplete Modality Disentangled Representation for Ophthalmic Disease Grading and Diagnosis (AAAI 2025 · similarity 0.84)
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.
