New Recurrent Contrastive Learning method boosts imbalanced medical image classification

Researchers have introduced Recurrent Contrastive Learning (RCL), a novel method designed to improve imbalanced medical image classification. RCL aims to expand the feature representation of underrepresented classes by reusing historical feature states throughout training. The approach utilizes DINOv3 with LoRA adapters as its backbone and incorporates a Temporal Memory Queue (TMQ) to maintain corpus-level features. This TMQ is used to create Temporal Anchors (TARs) that form a supportive field around tail classes, thereby enhancing their representation and improving separation from head classes. AI

IMPACT This research could lead to more accurate diagnoses in medical imaging by improving the handling of rare conditions.

RANK_REASON The cluster contains a research paper detailing a new method for medical image classification. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.CV →

• arXiv
• DINOv3
• Lora
• Recurrent Contrastive Learning
• Temporal Anchors
• Temporal Memory Queue

• paper
• other

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →