Three New arXiv Papers Advance Federated Learning Efficiency and Decision Systems

Three New arXiv Papers Advance Federated Learning Efficiency and Decision Systems

Three recent papers published on arXiv address different aspects of distributed machine learning and optimization systems.

Energy-Efficient Medical Imaging

According to arXiv paper 2512.03054v1, researchers propose an “energy-efficient federated learning via adaptive encoder freezing for MRI-to-CT conversion.” The paper notes that while Federated Learning (FL) “holds the potential to advance equality in health by enabling diverse institutions to collaboratively train deep learning models, even with limited data,” the “significant resource requirements of FL often exclude” certain participants. The research focuses on a “Green AI-Guided” approach to address these constraints.

Wireless Edge AI Integration

ArXiv paper 2512.03719v1 introduces “Over-the-Air Federated Learning (AirFL),” described as “an emerging paradigm that tightly integrates wireless signal processing and distributed machine learning to enable scalable AI at the network edge.” According to the abstract, the approach leverages “the superposition property of wireless signal” to enhance federated learning deployment.

Decision Support Optimization

ArXiv paper 2512.03861v1 addresses “scalable decision focused learning via online trainable surrogates.” The paper states that “decision support systems often rely on solving complex optimization problems that may require to estimate uncertain parameters beforehand,” and notes that “recent studies have shown how using traditionally trained estimators for this task can lead to suboptimal” results.