-
[pdf]
[supp]
[arXiv]
[bibtex]@InProceedings{Hu_2023_ICCV, author = {Hu, Erdong and Tang, Yuxin and Kyrillidis, Anastasios and Jermaine, Chris}, title = {Federated Learning Over Images: Vertical Decompositions and Pre-Trained Backbones Are Difficult to Beat}, booktitle = {Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV)}, month = {October}, year = {2023}, pages = {19385-19396} }
Federated Learning Over Images: Vertical Decompositions and Pre-Trained Backbones Are Difficult to Beat
Abstract
We carefully evaluate a number of algorithms for learning in a federated environment, and test their utility for a variety of image classification tasks. We consider many issues that have not been adequately considered before: whether learning over data sets that do not have diverse sets of images affects the results; whether to use a pre-trained feature extraction "backbone"; how to evaluate learner performance (we argue that classification accuracy is not enough), among others. Overall, across a wide variety of settings, we find that vertically decomposing a neural network seems to give the best results, and outperforms more standard reconciliation-used methods.
Related Material