FedCLAR: Federated Clustering for Personalized Sensor-Based Human Activity Recognition

Abstract

Sensor-based Human Activity Recognition (HAR) has been a hot topic in pervasive computing for several years mainly due to its applications in healthcare and well-being. Centralized supervised approaches reach very high recognition rates, but they incur privacy and scalability issues. Federated Learning (FL) has been recently proposed to mitigate these issues. Each subject only shares the weights of a personal model trained locally, instead of sharing data. A cloud server is in charge of aggregating the weights to generate a global model. However, since activity data is non-independently and identically distributed (non-IID), a single model may not be sufficiently accurate for a large number of diverse users. In this work, we propose FedCLAR, a novel federated clustering method for HAR. Based on the similarity of the local model updates, the cloud server in FedCLAR derives groups of users that exhibit similar ways of performing activities. For each group, FedCLAR uses a specialized global model to mitigate the non-IID problem. We evaluated FedCLAR on two well-known public datasets, showing that it outperforms state-of-the-art FL solutions.

Kavach: A Personalized Secure and Private Decentralized Learning Setup for Human Activity Recognition

“…Despite its reputation for computational complexity, homomorphic encryption (HE) has gained attention among researchers, resulting in the fusion of HAR and HE in several recent studies [39,13,14,28]. To mitigate the computational challenges associated with HE, Zhou et al [39] introduce partial homomorphic encryption as a viable solution, while Kim et al [13] employ HE specifically in the inference stage.…”

“…To mitigate the computational challenges associated with HE, Zhou et al [39] introduce partial homomorphic encryption as a viable solution, while Kim et al [13] employ HE specifically in the inference stage. While the majority of these works offer centralized HE solutions within their systems, Presotto et al [28] extend the use of HE to the federated context.…”

FedScrap: Layer-Wise Personalized Federated Learning for Scrap Detection

“…In [13], Wang et al use the parameters in the local training process as the cognitive basis and calculate Earth mover's distance to quantify the differences between different models. Presotto et al [32] proposed a federated clustering algorithm FedCLAR, which grouped clients based on the similarity of client models, so as to better identify and distinguish client data with different distributions. Yan et al [12] proposed ICFL, which can dynamically determine the cluster structure of clients during each training round and aggregate a personalized model for each cluster.…”

Federated Clustering and Semi-Supervised learning: A new partnership for personalized Human Activity Recognition

“…The main drawback of these approaches is that they consider a convex objective function, which is not suitable for complex HAR models based on deep learning. Recently, a few works proposed solutions based on Federated Clustering [37,38,13]. The goal of Federated Clustering is to create specialized global models (server-side) by grouping users that perform activities in a similar way.…”

“…In a recent work, we proposed FedCLAR to tackle this issue [13]. FedCLAR uses Federated Clustering to generate (server-side) specialized global models for groups of similar users.…”