FedClamp: An Algorithm for Identification of Anomalous Client in Federated Learning

Manzoor, Habib Ullah; Khan, Muhammad Bilal; Khan, Ahsan Raza; Ayaz, Fahad; Flynn, David; Imran, Muhammad Ali; Zoha, Ahmed

doi:10.1109/icecs202256217.2022.9970909

Cited by 11 publications

(10 citation statements)

References 21 publications

(20 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For instance, Qureshi suggests that in federated learning collaborative training for load forecasting, the server clusters the uploaded parameter weight features into two sets and identifies smaller sets as attackers before aggregating the parameters in the larger set directly with FedAvg [12]. Meanwhile, FedClamp proposes using Hidden Markov Models for poisoning model detection to identify attackers before model aggregation [13]. However, these approaches only detect potentially poisoning attacks, and the aggregation weights do not fully consider each participant's impact on the global model's weight.…”

Section: Related Workmentioning

confidence: 99%

“…The global model finally completes the aggregation of parameters, and the equation is shown in (13).…”

Section: Aggregation Of the Global Model Based On Distancementioning

confidence: 99%

“…For instance, the method that clustering algorithm to detect poisoning attacks fails to fully account for the quality of each participant's parameters during the aggregation process [12]. At the same time, FedClamp relies on a testing and validation method to ensure the reasonableness of uploaded parameters, which makes verification time longer [13].…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Fed‐SAD: A secure aggregation federated learning method for distributed short‐term load forecasting

Li,

Wu,

Wang

et al. 2023

IET Generation Trans & Dist

View full text Add to dashboard Cite

The distributed and privacy‐preserving attributes of fine‐grained smart grid data create obstacles to data sharing. As a result, federated learning emerges as an effective strategy for collaborative training in distributed load forecasting. However, poisoning attacks can interfere with training in the federated learning aggregation process, making it challenging to ensure the accuracy and safety of the global model in load forecasting. Therefore, the authors propose a secure aggregation federated learning method based on similarity and distance (Fed‐SAD) for distributed load forecasting. The server determines approximate global model parameters based on the similarity of the model parameters of the participants and aggregates the global model parameters using a distance‐based weighting method. Fed‐SAD effectively reduces the interference of poisoning attacks by securely aggregating models in short‐term load forecasting. Experimental results show that using the Fed‐SAD results in mean absolute percentage error (MAPE) reduction for certain participants by 19% on sign flipping attacks, MAPE reduction by 15% on additive noise attacks, and MAPE reduction by 4.5% without attack, compared to the Federated Average Aggregation algorithm. Furthermore, Fed‐SAD consistently maintains robustness and has excellent attack detection accuracy.

show abstract

Section: Related Workmentioning

confidence: 99%

“…The global model finally completes the aggregation of parameters, and the equation is shown in (13).…”

Section: Aggregation Of the Global Model Based On Distancementioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Fed‐SAD: A secure aggregation federated learning method for distributed short‐term load forecasting

Li,

Wu,

Wang

et al. 2023

IET Generation Trans & Dist

View full text Add to dashboard Cite

show abstract

“…Because the server does not have access to the clients’ data, there is the possibility that a client might not be sending proper updates to the server. This problem and its resolution have been discussed in detail in [ 12 ]. Similarly, data diversity is one of the biggest challenges in the FL environment.…”

Section: Introductionmentioning

confidence: 99%

FedBranched: Leveraging Federated Learning for Anomaly-Aware Load Forecasting in Energy Networks

Manzoor

Khan

Flynn

et al. 2023

Sensors

Self Cite

View full text Add to dashboard Cite

Increased demand for fast edge computation and privacy concerns have shifted researchers’ focus towards a type of distributed learning known as federated learning (FL). Recently, much research has been carried out on FL; however, a major challenge is the need to tackle the high diversity in different clients. Our research shows that using highly diverse data sets in FL can lead to low accuracy of some local models, which can be categorised as anomalous behaviour. In this paper, we present FedBranch, a clustering-based framework that uses probabilistic methods to create branches of clients and assigns their respective global models. Branching is performed using hidden Markov model clustering (HMM), and a round of branching depends on the diversity of the data. Clustering is performed on Euclidean distances of mean absolute percentage errors (MAPE) obtained from each client at the end of pre-defined communication rounds. The proposed framework was implemented on substation-level energy data with nine clients for short-term load forecasting using an artificial neural network (ANN). FedBranch took two clustering rounds and resulted in two different branches having individual global models. The results show a substantial increase in the average MAPE of all clients; the biggest improvement of 11.36% was observed in one client.

show abstract

“…There are a limited number of research methods that address secure aggregation based on federated learning in the context of distributed load forecasting, and these methods have certain limitations. For instance, the method that spectral clustering algorithm to detect poisoning attacks fails to fully account for the quality of each participant's parameters during the aggregation process [18], while FedClamp relies on a testing and validation method to ensure the reasonableness of uploaded parameters, which lengthens each round [14].…”

Section: Introductionmentioning

confidence: 99%

Fed-SAD:A secure aggregation federated learning method for distributed load forecasting

Wang

et al. 2023

Preprint

View full text Add to dashboard Cite

The distributed and privacy-preserving characteristics of fine-grained smart grid data hinder data sharing, making federated learning an attractive approach for collaborative training among data owners with similar load patterns. However, malicious models can interfere with training in the federated learning aggregation process, making it difficult to ensure the accuracy and safety of the central model in load forecasting. Therefore, we propose a secure aggregation federated learning method for distributed load forecasting based on similarity and distance (Fed-SAD), which effectively eliminates the interference of malicious models by securely aggregating models, thereby ensuring accurate and safe distributed scenario prediction. Experimental results demonstrate that Fed-SAD maintains high accuracy and robustness in both the presence and absence of malicious models, while maintaining data and model security.

show abstract

FedClamp: An Algorithm for Identification of Anomalous Client in Federated Learning

Abstract: There may be differences between this version and the published version. You are advised to consult the published version if you wish to cite from it.

Cited by 11 publications

References 21 publications

Fed‐SAD: A secure aggregation federated learning method for distributed short‐term load forecasting

Fed‐SAD: A secure aggregation federated learning method for distributed short‐term load forecasting

FedBranched: Leveraging Federated Learning for Anomaly-Aware Load Forecasting in Energy Networks

Fed-SAD:A secure aggregation federated learning method for distributed load forecasting

Contact Info

Product

Resources

About