Exploiting Unlabeled Data in Smart Cities using Federated Edge Learning

Abstract

Privacy concerns are considered one of the main challenges in smart cities as sharing sensitive data induces threatening problems in people's lives. Federated learning has emerged as an effective technique to avoid privacy infringement as well as increase the utilization of the data. However, there is a scarcity in the amount of labeled data and an abundance of unlabeled data collected in smart cities; hence there is a necessity to utilize semi-supervised learning. In this paper, we present the primary design aspects for enabling federated learning at the edge networks taking into account the problem of unlabeled data. We propose a semi-supervised federated edge learning method called FedSem that exploits unlabeled data in real-time. FedSem algorithm is divided into two phases. The first phase trains a global model using only the labeled data. In the second phase, Fedsem injects unlabeled data into the learning process using the pseudo labeling technique and the model developed in the first phase to improve the learning performance. We carried out several experiments using the traffic signs dataset as a case study. Our results show that FedSem can achieve accuracy by up to 8% by utilizing the unlabeled data in the learning process.

Federated Learning for Internet of Things: A Comprehensive Survey

“…The use of massive data from multiple vehicles and the large computational capability of all participant help provide better traffic prediction outcomes, which cannot be achieved by using centralized ML techniques with less dataset and limited computation. FL has been exploited to provide distributed AI functions for decentralized smart city applications such as intelligent smart city data management [48]. In this context, FL is helpful to structure data streams from ubiquitous IoT devices that work as FL clients for performing local learning without sharing their data to external third-parties.…”

“…1) FL for Data Management: With its decentralized and privacy-preserved nature, FL has been exploited to provide distributed AI functions for large-scale intelligent data management systems in smart cities. For example, a semi-supervised FL method called FedSem is introduced in [48] to provide distributed processing for unlabeled data in smart cities. To evaluate the usefulness of FL in a smart city, a prototype with smart vehicles is considered where each vehicle learns a DNN model based on traffic sign image datasets.…”

“…However, most proposed AI-based smart city solutions rely on a centralized learning architecture on a data center, such as a cloud server, which is obviously not scalable to the rapid expansion of smart devices in smart cities. FL offers more attractive features for enabling decentralized smart city applications with high privacy levels and low communication delays [48]. FL is also important for structuring data streams from ubiquitous IoT devices that act as FL clients for performing local learning without sharing their data with external third-parties [153].…”

Less is More: A privacy-respecting Android malware classifier using federated learning

“…Semi-supervised FL has already been proposed by [2] to take advantage of the abundant unlabeled data in the smart city context. Contrary to LiM, they assume a subset of clients have labeled samples, and use them to train a classifier that will provide the missing labels to retrain another local model.…”

Rebirth of Distributed AI—A Review of eHealth Research

“…Smart cities aim at providing robust solutions to crucial societal challenges related to transportation, health, environment, education, and security [ 1 ]. Smart cities are expected to deploy massive Internet-of-Things (IoT) related devices and applications.…”