FACT: Federated Adaptive Cross Training

Federated Learning (FL) facilitates distributed model development to aggregate multiple confidential data sources. The information transfer among clients can be compromised by distributional differences, i.e., by non-i.i.d. data. A particularly challenging scenario is the federated model adaptation to a target client without access to annotated data. We propose Federated Adaptive Cross Training (FACT), which uses the implicit domain differences between source clients to identify domain shifts in the target domain. In each round of FL, FACT cross initializes a pair of source clients to generate domain specialized representations which are then used to enforce a domain invariant data representation. We empirically show that FACT outperforms both state-of-the-art federated and non-federated models on three popular multi-source-single-target benchmarks, and achieves highly competitive performance on single-source-single-target experiments. We further study FACT’s behavior with respect to communication restrictions and the number of participating clients. • FACT uses unsupervised domain adaptation to integrate distributed data across domains. • FACT preserves clients’ privacy and facilitates federated learning tasks. • FACT outperformed state-of-the-art models in three popular benchmark datasets.