Search papers, labs, and topics across Lattice.
This work was partially supported by the Wallenberg AI, Autonomous Systems and Software Program (WASP) funded by the Knut and Alice Wallenberg Foundation via the WASP NEST project “Intelligent Cloud Robotics for Real-Time Manipulation at Scale.” The computations and data handling essential to our research were enabled by the supercomputing resource Berzelius provided by the National Supercomputer Centre at Linköping University and the gracious support of the Knut and Alice Wallenberg Foundation
2
0
3
2
Exposing intermediate representations in federated split learning can leak client data, but a new defense cuts structural similarity in reconstructed images by up to 40% without hurting model performance.
By synthesizing data with a diffusion model and selectively retaining informative samples, OSI-FL achieves state-of-the-art performance in class-incremental and domain-incremental federated learning scenarios while minimizing communication costs.
