LOSCAR-SGD combines local updates, sparse model averaging, and communication-computation overlap with a delay-corrected merge rule, providing convergence rates for smooth non-convex objectives under worker heterogeneity.
hub Canonical reference
Communication-Efficient Learning of Deep Networks from Decentralized Data
Canonical reference. 80% of citing Pith papers cite this work as background.
30
Pith papers citing it
Background
80% of classified citations
hub tools
citation-role summary
background 4
method 1
citation-polarity summary
representative citing papers
PMF-CL derives Pareto-minimal-forgetting algorithms for linear/basis-function regression and quadratic-bounded losses like logistic regression, achieving static O(d²) memory for d-parameter models.
Introduces the CULT threat model with four circuit-level attacks on quantum federated learning and shows they degrade accuracy on MNIST and CIFAR-10 even when defenses like Krum are used.
Ringmaster LMO extends delay-thresholding from ASGD to LMO-based momentum updates, providing convergence guarantees under (L0, L1)-smoothness and time-complexity bounds that recover optimal rates in the Euclidean case.
Real 5G testbed experiments show consistent stragglers in 70% of federated learning trials due to communication delays, challenging common wireless FL assumptions.
Prefix-tuning matches or exceeds fine-tuning on NLG tasks by optimizing a continuous prefix using 0.1% of parameters while keeping the LM frozen.
Algorithm for group distributionally robust linear regression using block Lewis weights to achieve (1+ε) optimality in Õ(min{rank(A), m}^{1/3} ε^{-2/3}) linear-system solves.
FLKit is a new toolkit with four lifecycle stages, eleven role-specific entry points, a glossary, FL Story template, and tool directory to support federated learning projects in health and life sciences.
A multimodal graph learning method for V2X beam alignment cuts overhead by over 90% and outperforms prior federated learning baselines under label and modality imbalance.
Decoupled DiLoCo enables asynchronous distributed pre-training with zero global downtime under simulated failures while preserving competitive performance on text and vision tasks.
Random Network Distillation enables pre-training discovery of client clusters in federated learning via local novelty signals, supporting autonomous grouping under non-IID data without a priori cluster count.
FPLIER performs federated PLIER training via secure aggregation that is algebraically equivalent to centralized training, with membership-inference risk shown to decrease as the rank of the expression matrix increases.
A federated intrusion detection method combines hybrid Naive Bayes classifiers as a mixture of Gaussians and uses a governance-derived Institutional Coherence Index to regularize server-side weights via Nelder-Mead optimization, reporting F1 gains over size-proportional averaging on three datasets.
BiFedKD improves ECG classification accuracy by 3.52% and Macro-F1 by 9.93% on MIT-BIH while cutting communication overhead 40% and computation cost 71.7% versus baseline federated methods.
Rennala MVR improves time complexity over Rennala SGD for smooth nonconvex stochastic optimization in heterogeneous parallel systems under a mean-squared smoothness assumption.
Experiments on real industrial time series show that partial model sharing improves diffusion model performance in bandwidth-limited non-IID settings, while full sharing stabilizes GAN training but offers less robustness than VAE or DDPM alternatives.
Federated learning on 310 CT scans from two centers yields pediatric OAR segmentation models with better cross-center robustness than local models for nine evaluated structures.
SplitFT adapts cut-layer selection and reduces LoRA rank per client in federated split learning to improve efficiency and performance when fine-tuning LLMs on heterogeneous devices and data.
CPPDD is a new consensus-based protocol for privacy-preserving multi-client data sharing that achieves unanimous-release confidentiality, linear scalability, and high-probability malicious deviation detection.
FoggyTrust is a hierarchical extension of FLTrust that localizes trust computation to fog nodes and combines it with heterogeneity-aware optimizers, reporting over 50% gains on CIFAR-10 under Krum and Trim attacks.
SwarmHarness is a proposed decentralized protocol for compute sharing among AI agents via DHT registry, load-aware routing, and credit incentives that penalize non-contributors.
Life-logging video streams create an inevitable privacy-utility trade-off that is a foundational challenge for always-on AI systems.
FedProx outperforms FedAvg for deeper models under data heterogeneity, BSP reaches near-centralized accuracy at high communication cost, and LeNet gives the best accuracy-communication trade-off on the UC Merced dataset.
Iterative joint learning-optimization framework with convergent algorithms for pseudoconvex objectives in operational decision systems.
citing papers explorer
-
On the Tradeoffs of On-Device Generative Models in Federated Predictive Maintenance Systems
Experiments on real industrial time series show that partial model sharing improves diffusion model performance in bandwidth-limited non-IID settings, while full sharing stabilizes GAN training but offers less robustness than VAE or DDPM alternatives.