… privacy guarantees by incorporating user sampling to the private … with user sampling, where
users are sampled uniformly or … proposed sampling schemes, in which each user transmits …

… Our results suggest that there is significant loss from selecting users, as the difference
between Random Sampling and Active Sampling is much larger for server-side learning. …

… However, the performance of federated learning cannot be … data sampling and user selection
is studied in this paper. First, to capture the key features of deploying federated learning in …

… Abstract—Federated learning (FL) algorithms usually sample … on unbiased client sampling,
eg, sampling uniformly at … secure aggregation for federated learning on user-held data,…

… experiments with homogeneous data and convex learning objective, we use a synthetic
dataset and a logistic regression task. We generate this data by taking 104 samples Xi ∈ R100 …

… can be a primary bottleneck in federated learning (FL). In this work, … SGD (DSGD) and
Federated Averaging (FedAvg), the … We use vanilla SGD optimizers with constant step sizes, with …

… MD) and Uniform sampling, two default unbiased client sampling … MD sampling should be
used as default sampling scheme, … ratio during the learning process, while Uniform sampling is …

… between server and clients in federated learning (FL). Current sampling approaches in FL
… We use N = 100, m = 10, and respective learning rate for each dataset lr = {0.05, 0.05, 0.05, …

… To further protect the existence of each sample, we use a widely adopted randomized response
mechanism [32] to generate a noisy representation h(φk,i) of each projection vector h(φk,…

… for federated learning (FL) on cell-free massive multiple-input multiple-output networks. To
show how these approaches work, we consider a general FL framework with UE sampling, …