Data-Quality Based Scheduling for Federated Edge Learning

Abstract

FEderated Edge Learning (FEEL) has emerged as a leading technique for privacy-preserving distributed training in wireless edge networks, where edge devices collaboratively train machine learning (ML) models with the orchestration of a server. However, due to frequent communication, FEEL needs to be adapted to the limited communication bandwidth. Furthermore, the statistical heterogeneity of local datasets' distributions, and the uncertainty about the data quality pose important challenges to the training's convergence. Therefore, a meticulous selection of the participating devices and an analogous bandwidth allocation are necessary. In this paper, we propose a data-quality based scheduling (DQS) algorithm for FEEL. DQS prioritizes reliable devices with rich and diverse datasets. In this paper, we define the different components of the learning algorithm and the data-quality evaluation. Then, we formulate the device selection and the bandwidth allocation problem. Finally, we present our DQS algorithm for FEEL, and we evaluate it in different data poisoning scenarios.

Empowering Patient Similarity Networks through Innovative Data-Quality-Aware Federated Profiling

“…Learning [51] Uses DQ aspects for scheduling edges in federated edge learning, prioritizes reliable devices with diverse datasets.…”

“…Federated edge learning’s privacy-preserving aspect makes it the future of wireless edge network training. In [ 51 ], data-quality aspects are used to schedule edges for collaborative training. First, the learning algorithm’s components, dataset diversity, and edge node reputation are defined.…”

Toward Trustworthy and Responsible Autonomous Drones in Future Smart Cities

“…Data bias and drift detection Drones that are deployed should continue learning over time to improve their operation , e.g., by taking advantage of federated learning [7]. Optimally this process should integrate data from drones operating in different parts of the environment as this helps improving the generality and robustness of the AI models.…”

“…The drone records video footage of disposed litter which is pre-processed and analyzed to identify litter [3]. To attack the model, we poison the data using blurring [7].…”

“…Explainable AI (XAI) methods have recently emerged as a mechanism that can help to understand the behaviour of AI models and the factors affecting them. For example, XAI methods can be used to evaluate the quality of the data used for training [7] or to assess the suitability of the internal model structure [8]. While these methods provide a foundation for analyzing the trustworthiness of the AI models integrated onto drones, they alone are insufficient as it is difficult to distinguish between multiple factors that have a similar effect on the model.…”