Federated learning (FL) has been intensively investigated in terms of communication efficiency, privacy, and fairness. However, efficient annotation, which is a pain point in real-world FL applications, is less studied. In this project, we propose to apply active learning (AL) and sampling strategy into the FL framework to reduce the annotation workload. We expect that the AL and FL can improve the performance of each other complementarily. In our proposed federated active learning (F-AL) method, the clients collaboratively implement the AL to obtain the instances which are considered as informative to FL in a distributed optimization manner. We compare the test accuracies of the global FL models using the conventional random sampling strategy, client-level separate AL (S-AL), and the proposed F-AL. We empirically demonstrate that the F-AL outperforms baseline methods in image classification tasks.
This paper concerns the feasibility of full-duplex large-scale multiple-input-multiple-output (MIMO) cellular systems. We first propose a pilot transmission scheme and assess its performance, specifically the ergodic sum-rate. The proposed scheme -the simultaneous pilot transmission (SPT) -enables to reduce pilot overhead, where the pilot overhead depends on the number of antennas at the base station (BS), since the self-interference channel has to be estimated. We consider two multicell scenarios-cooperative and non-cooperative multicell systems-, and derive the analytic model of the ergodic achievable sumrate for cell-boundary users. The model is derived by applying a simple linear filter, i.e., matched filter or zero-forcing filter, to the BS. In the analytic model, we also consider large-scale fading, pilot contamination, transmitter noise and receiver distortion. Exploiting the derived analytic model, the feasibility of full-duplex large-scale MIMO systems is shown with respect to system parameters.In the end, we confirm that our analytic model matches well the numerical results and the SPT has advantages over other pilot transmission methods.
Index TermsFull-duplex, large-scale MIMO, massive MIMO, cloud radio access networks, and channel estimation.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.