Congratulations to Ziqi and Qiang for CRAFT getting accepted at NeurIPS đ The paper not only shows how useful conflict resolution is in the context of federated learning, it also makes several important contributions to the ConFIG optimizer. Most importantly:
Most importantly:
- the insight that conflict resolution can be performed on different levels, and that a finer-grained approach (e.g. layer wise) typically yields benefits,
- that anchoring the updates to a target direction, here from previous steps, provides increased stability.
These changes provide improvements for the general usage of ConFIG in different contexts, and help to further establish the importance of conflict resolution when facing multiple learning objectives.
Pre-print on arXiv: https://arxiv.org/abs/2605.21317
Code on GitHub: https://github.com/tum-pbs/CRAFT
Full paper abstract: The aggregation of conflicting client updates remains a fundamental bottleneck in federated learning (FL) under heterogeneous data distributions. Naive averaging can produce a global update that improves the global objective while conflicting with specific clients, causing degradation for those clients. In this work, we propose CRAFT (Conflict-Resolved Aggregation for Federated Training), a new aggregation framework that treats the global update as a geometric correction problem. We formulate aggregation as finding the update closest to a reference direction while satisfying conflict-free alignment constraints. We derive a closed-form expression for the constrained optimization problem, avoiding the computational overhead of iterative solvers. Furthermore, we use a layer-wise adaptation to address conflicts at varying feature granularities. We provide a theoretical analysis showing that CRAFT promotes a common-descent structure and mitigates conflicts through its projection geometry. Extensive experiments on heterogeneous benchmarks demonstrate that CRAFT improves the accuracy of the global model while reducing performance disparity across clients compared with state-of-the-art baselines.

