Sparse Uncertainty-Informed Sampling from Federated Streaming Data
Abstract: We present a numerically robust, computationally efficient approach for non-I.I.D. data stream sampling in federated client systems, where resources are limited and labeled data for local model adaptation is sparse and expensive. The proposed method identifies relevant stream observations to optimize the underlying client model, given a local labeling budget, and performs instantaneous labeling decisions without relying on any memory buffering strategies. Our experiments show enhanced training batch diversity and an improved numerical robustness of the proposal compared to existing strategies over large-scale data streams, making our approach an effective and convenient solution in FL environments.
- B. McMahan et al. Communication-efficient learning of deep networks from decentralized data. In Proc. 20th AISTAT, volume 54, pages 1273–1282, April 2017.
- N. Rieke. The future of digital health with federated learn. Digital Medicine, 3, 12 2020.
- G. Long et al. Federated learning for open banking. In Fed. Learn. - Privacy and Incentive, volume 12500 of LNCS, pages 240–254. Springer, 2020.
- H. Zhang et al. End-to-end federated learning for autonomous driving vehicles. In 2021 IJCNN, pages 1–8, 2021.
- M. Heusinger. Learning with high dimensional data and preprocessing in non-stationary environments. PhD thesis, Bielefeld University, Germany, 2023.
- O. Marfoq et al. Federated learning for data streams. In Proc. 26th AISTAT, volume 206, pages 8889–8924. PMLR, 25–27 Apr 2023.
- B. Settles. Active learning literature survey. CS Tech. Rep. 1648, WISC, 2009.
- A. Saran et al. Streaming active learning with deep neural networks. In Proc. 40th ICML, ICML’23. JMLR.org, 2023.
- M. Woodbury. Inverting modified matrices. In Mem. Rept. 42, page 4. Princeton, 1950.
- M. Seeger. Low rank updates for the cholesky decomposition. 2004.
- M. Röder et al. Crossing Domain Borders with Federated Few-Shot Adaptation. In Proc. 13th ICPRAM, pages 511–521, 2024.
- G. Golub et al. Matrix Computations, pages 338–341. JHU Press, 20134.
- A. Schulz et al. Deepview: Visualizing classification boundaries of deep neural networks as scatter plots using discriminative dimensionality reduction. In Proc., IJCAI 2020, pages 2305–2311. ijcai.org, 2020.
Paper Prompts
Sign up for free to create and run prompts on this paper using GPT-5.
Top Community Prompts
Collections
Sign up for free to add this paper to one or more collections.