The field of federated learning is moving towards increased privacy preservation and efficiency, with a focus on capturing spatial relationships and handling heterogeneous data. Researchers are exploring novel frameworks and algorithms that enable collaborative learning while minimizing resource demands and maintaining high accuracy. Noteworthy papers in this area include:

• A study proposing FLLL3M, a privacy-preserving framework for Next-Location Prediction that achieves state-of-the-art results on multiple datasets while reducing parameters and memory usage.
• A paper introducing a lightweight graph-aware federated learning approach that captures spatial relationships effectively while remaining computationally efficient.
• A work proposing two new federated learning algorithms, Fed-Cyclic and Fed-Star, that perform better than existing baselines on a newly introduced image classification dataset. These advancements have the potential to significantly impact various applications, including traffic prediction and image classification, by enabling more accurate and efficient models while preserving data privacy.