Edge-Assisted Democratized Learning Toward Federated Analytics

A recent take toward federated analytics (FA), which allows analytical insights of distributed data sets, reuses the federated learning (FL) infrastructure to evaluate the summary of model performances across the training devices. However, the current realization of FL adopts single server-multiple client architecture with limited scope for FA, which often results in learning models with poor generalization, i.e., an ability to handle new/unseen data, for real-world applications. Moreover, a hierarchical FL structure with distributed computing platforms demonstrates incoherent model performances at different aggregation levels. Therefore, we need to design a robust learning mechanism than the FL that 1) unleashes a viable infrastructure for FA and 2) trains learning models with better generalization capability. In this work, we adopt the novel democratized learning (Dem-AI) principles and designs to meet these objectives. First, we show the hierarchical learning structure of the proposed edge-assisted Dem-AI mechanism, namely Edge-DemLearn, as a practical framework to empower generalization capability in support of FA. Second, we validate Edge-DemLearn as a flexible model training mechanism to build a distributed control and aggregation methodology in regions by leveraging the distributed computing infrastructure. The distributed edge computing servers construct regional models, minimize the communication loads, and ensure distributed data analytic application’s scalability. To that end, we adhere to a near-optimal two-sided many-to-one matching approach to handle the combinatorial constraints in Edge-DemLearn and solve it for fast knowledge acquisition with optimization of resource allocation and associations between multiple servers and devices. Extensive simulation results on real data sets demonstrate the effectiveness of the proposed methods.

I. Introduction

Recently, the Google AI team published an article on leveraging computing mechanism of the distributed learning model training infrastructure to facilitate data analytics, namely federated analytics (FA) [1]. FA allows data scientists to derive analytical insights of distributed data sets without the need of moving data to a central computing entity. This concept gathered keen attention for a new approach to data science, and interestingly, at the time when the centralized repositories are termed “vulnerable” toward privacy for data collection, and the era of distributed computing and storage is prominent. Besides, it means that apart from considering the distributed model training processes for improving model accuracy, we can exploit such collaboration architecture to evaluate the quality of the trained model at the user-level perspectives, i.e., the model performance at the user’s end. Hence, without the learning part, we can reuse the computing scheme of the learning architecture to perform statistical analysis on local data that may lead to building better products. To elaborate this idea further, consider an example of a prediction model where the developer would be interested in finding popular content to store in a shared regional database without breaking into user’s historical content usage data. An intuitive answer to this question would be to find the frequently requested content at first, which is best done with FA. This is similar to the Now Playing feature on Google’s Pixel phones for managing regional song database [1] to show users songs playing around them. In addition, leveraging distributed multiaccess edge computing (MEC) servers further allows fast knowledge acquisition to serve user’s requests, limit privacy leakages, and improve the on-the-fly learning process. Furthermore, this also brings FA closer to where the data is collected, i.e., at the one-hop proximity of user devices. In such scenarios, the developer’s objective would be to improve the model’s accuracy and, concurrently, enhance the generalization performance of the model at the user level for maintaining regional databases using FA.