Geometry-Aware Uncertainty Coresets for Robust Visual In-Context Learning in Histopathology

📰 ArXiv cs.AI

Learn to improve visual in-context learning in histopathology using geometry-aware uncertainty coresets, enhancing robustness and accuracy

advanced Published 19 May 2026

Action Steps

Apply geometry-aware uncertainty coresets to select representative image-text pairs for in-context learning
Use vision-language models (VLMs) to couple visual perception with clinical reasoning in histopathology
Fine-tune VLMs on scarce, expert-annotated pathology data using the selected coresets
Evaluate the performance of the VLMs using metrics such as accuracy and robustness
Compare the results with traditional in-context learning methods to demonstrate the effectiveness of geometry-aware uncertainty coresets

Who Needs to Know This

This research benefits teams working on computational histopathology, particularly those using vision-language models (VLMs) for image-text analysis, as it provides a method to improve the robustness of in-context learning

Key Insight

💡 Geometry-aware uncertainty coresets can improve the robustness of in-context learning in histopathology by selecting representative image-text pairs

Key Takeaways

Learn to improve visual in-context learning in histopathology using geometry-aware uncertainty coresets, enhancing robustness and accuracy

Full Article

Title: Geometry-Aware Uncertainty Coresets for Robust Visual In-Context Learning in Histopathology

Abstract:
arXiv:2605.18419v1 Announce Type: cross Abstract: Vision-language models (VLMs) can couple visual perception with open-ended clinical reasoning, making them attractive for computational histopathology. However, fine-tuning billions of parameters on scarce, expert-annotated pathology data is prohibitive, while in-context learning (ICL), which conditions the VLM on demonstrative image-text pairs without parameter updates, suffers from high sensitivity to which examples are selected and how the que

Read full paper → ← Back to Reads