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

Presotto, Riccardo; Civitarese, Gabriele; Bettini, Cláudio

doi:10.1016/j.pmcj.2022.101726

Cited by 9 publications

(4 citation statements)

References 54 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Personalized human activity recognition (HAR) integrates federated clustering with semi-supervised learning to capture the heterogeneity of data. The suggestion method dealt with the issue of unlabeled data by creating pseudolabels [48]. In [49], dynamic clusters based on the cosine transform and affinity index were used to address the unbalanced data volume of clients and skewed data, and the authors were able to enhance the accuracy of the model by 20%.…”

Section: Hybrid Skewmentioning

confidence: 99%

“…Another paper [89] introduced the first federated transfer learning system for real Parkinson's disease auxiliary and wearable healthcare activity identification experiments. One author designed a framework for classifying human activity such as walking, sitting, standing, and stretching; the system can mitigate data heterogeneity and unlabeled data [48]. In the same way, another author developed [32] a secure framework for personal HAR reinforcement in CPSS and fixed the problem of patients not having enough activity data.…”

Section: ) Covid-19mentioning

confidence: 99%

See 1 more Smart Citation

A Survey of Federated Learning From Data Perspective in the Healthcare Domain: Challenges, Methods, and Future Directions

et al. 2023

View full text Add to dashboard Cite

Recent advances in deep learning (DL) have shown that data-driven insights can be used in smart healthcare applications to improve the quality of life for patients. DL needs more data and diversity to build a more accurate system. To satisfy these requirements, more data need to be pooled at the centralized server to train the model deeply, but the process of pooling faces privacy and regulatory challenges. To settle them, the concept of sharing model learning rather than sharing data through federated learning (FL) is proposed. FL creates a more reliable system without transferring data to the server, resulting in the right system with stronger security and access rights to data that protect privacy. This research aims to (1) provide a literature review and an in-depth study on the roles of FL in the fields of healthcare; (2) highlight the effectiveness of current challenges facing standardized FL, including statistical data heterogeneity, privacy and security concerns, expensive communications, limited resources, and efficiency; and (3) present lists of open research challenges and recommendations for future FL for the academic and industrial sectors in telemedicine and remote healthcare applications. An extensive review of the literature on FL from a data-centric perspective was conducted. We searched the Science Direct, IEEE Xplore, and PubMed databases for publications published between January 2018 and February 2023. A new crossover matching between the approaches that solve or mitigate all types of skewed data has been proposed to open up opportunities to other researchers. In addition, a list of various applications was organized by learning application task types such as prediction, diagnosis, and classification. We think that this study can serve as a helpful manual for academics and industry professionals, giving them guidance and important directions for future studies.

show abstract

Section: Hybrid Skewmentioning

confidence: 99%

Section: ) Covid-19mentioning

confidence: 99%

A Survey of Federated Learning From Data Perspective in the Healthcare Domain: Challenges, Methods, and Future Directions

et al. 2023

View full text Add to dashboard Cite

show abstract

“…Once the in-memory buffer is full, we will perform semi-supervised learning to acquire labels and update the classifier. Semi-supervised learning has long been applied in scenarios with scarce annotation [54]. The goal is to achieve high accuracy on label propagation and at the same time to require as few real labels as possible.…”

Section: Semi-supervised Learningmentioning

confidence: 99%

Online Continual Learning for Human Activity Recognition

Schiemer

Fang

Dobson

et al. 2023

Preprint

View full text Add to dashboard Cite

“…Specifically, it assumes that the central server has a small amount of labeled data to train an initial global model, which is then used to obtain pseudo-labels for local unlabeled samples that meet certain entropy conditions (i.e., below a selected entropy threshold). Presotto et al proposed SS-FedCLAR, a FL framework that combines Federated Clustering and Semi-Supervised Learning to mitigate both the non-IID and data scarcity problems for Personalized Sensor-Based Human Activity Recognition [33]. Specifically for semi-supervised learning, active learning and label propagation were combined to train local models without labeled data in clients.…”

Section: Related Workmentioning

confidence: 99%

Semi-supervised Graph Instance Transformer for Mental Health Inference

Dong

Tang

Cai

et al. 2021

2021 20th IEEE International Conference on Machine Learning and Applications (ICMLA)

View full text Add to dashboard Cite

Mobile sensing appears as a promising solution for health inference problem (e.g., influenza-like symptom recognition) by leveraging diverse smart sensors to capture fine-grained information about human behaviors and ambient contexts. Centralized training of machine learning models can place mobile users' sensitive information under privacy risks due to data breach and misexploitation. Federated Learning (FL) enables mobile devices to collaboratively learn global models without the exposure of local private data. However, there are challenges of on-device FL deployment using mobile sensing: 1) long-term and continuously collected mobile sensing data may exhibit domain shifts as sensing objects (e.g. humans) have varying behaviors as a result of internal and/or external stimulus; 2) model retraining using all available data may increase computation and memory burden; and 3) the sparsity of annotated crowd-sourced data causes supervised FL to lack robustness. In this work, we propose FedMobile, an incremental semi-supervised federated learning algorithm, to train models semi-supervisedly and incrementally in a decentralized online fashion. We evaluate FedMobile using a real-world mobile sensing dataset for influenza-like symptom recognition. Our empirical results show that FedMobile-trained models achieve the best results in comparison to the selected baseline methods.

show abstract

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

Cited by 9 publications

References 54 publications

A Survey of Federated Learning From Data Perspective in the Healthcare Domain: Challenges, Methods, and Future Directions

A Survey of Federated Learning From Data Perspective in the Healthcare Domain: Challenges, Methods, and Future Directions

Online Continual Learning for Human Activity Recognition

Semi-supervised Graph Instance Transformer for Mental Health Inference

Contact Info

Product

Resources

About