Green AI strategy cuts federated learning energy use by 23% for medical imaging

Researchers have developed a novel Green AI-guided strategy for federated learning that significantly reduces energy consumption and computational load during MRI-to-CT conversion tasks. This adaptive layer-freezing method selectively freezes encoder weights, leading to up to a 23% reduction in training time, energy use, and CO2 emissions without compromising model performance. The approach aims to enhance equity in healthcare by making collaborative deep learning more accessible to institutions with limited computational resources, promoting sustainability alongside advancements in AI-driven healthcare. AI

IMPACT Enables more equitable access to collaborative AI training in healthcare by reducing computational costs.

RANK_REASON Academic paper detailing a new methodology for energy-efficient federated learning. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.AI →

• Ciro Benito Raggio
• CodeCarbon
• federated learning
• Green AI
• mean absolute error
• MRI-to-CT conversion

• paper
• infra
• other

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