MedCLIP: Contrastive Learning from Unpaired Medical Images and Text

Computer Science

• Contrastive Learning 100%
• Training Data 50%
• False Negative 50%
• Zero-Shot Learning 50%
• Supervised Classification 25%
• Medical Knowledge 25%
• Information Retrieval 25%

Keyphrases

• MedCLIP 100%
• InfoNCE Loss 33%
• Multimodal Contrastive Learning 33%
• Matching Loss 33%
• Pre-Training Data 33%
• Image-Text Dataset 33%