Comprehensive Analysis Over Centralized and Federated Learning-Based Anomaly Detection in Networks with Explainable AI (XAI)

Yasintha Rumesh; Thulitha Theekshana Senevirathna; Pawani Porambage; Madhusanka Liyanage; Mika Ylianttila

doi:10.1109/ICC45041.2023.10278845

Comprehensive Analysis Over Centralized and Federated Learning-Based Anomaly Detection in Networks with Explainable AI (XAI)

Yasintha Rumesh
, Thulitha Theekshana Senevirathna
, Pawani Porambage
, Madhusanka Liyanage
, Mika Ylianttila

BA6402 Future communication networks

Research output: Chapter in Book/Report/Conference proceeding › Conference article in proceedings › Scientific › peer-review

4 Citations (Scopus)

Abstract

Many forms of machine learning (ML) and artificial intelligence (AI) techniques are adopted in communication networks to perform all optimizations, security management, and decision-making tasks. Instead of using conventional blackbox models, the tendency is to use explainable ML models that provide transparency and accountability. Moreover, Federate Learning (FL) type ML models are becoming more popular than the typical Centralized Learning (CL) models due to the distributed nature of the networks and security privacy concerns. Therefore, it is very timely to research how to find the explainability using Explainable AI (XAI) in different ML models. This paper comprehensively analyzes using XAI in CL and FL-based anomaly detection in networks. We use a deep neural network as the black-box model with two data sets, UNSW-NB15 and NSLKDD, and SHapley Additive exPlanations (SHAP) as the XAI model. We demonstrate that the FL explanation differs from CL with the client anomaly percentage.

Original language	English
Title of host publication	ICC 2023 - IEEE International Conference on Communications
Editors	Michele Zorzi, Meixia Tao, Walid Saad
Publisher	IEEE Institute of Electrical and Electronic Engineers
Pages	4853-4859
ISBN (Electronic)	978-1-5386-7462-8
DOIs	https://doi.org/10.1109/ICC45041.2023.10278845
Publication status	Published - 2023
MoE publication type	A4 Article in a conference publication
Event	IEEE International Conference on Communications, ICC 2023 - Rome, Italy Duration: 28 May 2023 → 1 Jun 2023

Conference

Conference	IEEE International Conference on Communications, ICC 2023
Country/Territory	Italy
City	Rome
Period	28/05/23 → 1/06/23

Funding

This work is partly supported by VTT Technical Research Centre of Finland and by Business Finland in SUNSET-6G and DROLO, European Union in SPATIAL (Grant No: 101021808), Academy of Finland in 6Genesis (grant no. 318927) and Science Foundation Ireland under CONNECT phase 2 (Grant no. 13/RC/2077_P2) projects.

Keywords

6G
Privacy
Explainable AI
Federated Learning
Centralized Learning
Security

Access to Document

10.1109/ICC45041.2023.10278845

Cite this

Rumesh, Y., Senevirathna, T. T., Porambage, P., Liyanage, M., & Ylianttila, M. (2023). Comprehensive Analysis Over Centralized and Federated Learning-Based Anomaly Detection in Networks with Explainable AI (XAI). In M. Zorzi, M. Tao, & W. Saad (Eds.), ICC 2023 - IEEE International Conference on Communications (pp. 4853-4859). IEEE Institute of Electrical and Electronic Engineers. https://doi.org/10.1109/ICC45041.2023.10278845

Rumesh, Yasintha ; Senevirathna, Thulitha Theekshana ; Porambage, Pawani et al. / Comprehensive Analysis Over Centralized and Federated Learning-Based Anomaly Detection in Networks with Explainable AI (XAI). ICC 2023 - IEEE International Conference on Communications. editor / Michele Zorzi ; Meixia Tao ; Walid Saad. IEEE Institute of Electrical and Electronic Engineers, 2023. pp. 4853-4859

@inproceedings{828f6ce8abac4bf78612a011486ec97b,

title = "Comprehensive Analysis Over Centralized and Federated Learning-Based Anomaly Detection in Networks with Explainable AI (XAI)",

abstract = "Many forms of machine learning (ML) and artificial intelligence (AI) techniques are adopted in communication networks to perform all optimizations, security management, and decision-making tasks. Instead of using conventional blackbox models, the tendency is to use explainable ML models that provide transparency and accountability. Moreover, Federate Learning (FL) type ML models are becoming more popular than the typical Centralized Learning (CL) models due to the distributed nature of the networks and security privacy concerns. Therefore, it is very timely to research how to find the explainability using Explainable AI (XAI) in different ML models. This paper comprehensively analyzes using XAI in CL and FL-based anomaly detection in networks. We use a deep neural network as the black-box model with two data sets, UNSW-NB15 and NSLKDD, and SHapley Additive exPlanations (SHAP) as the XAI model. We demonstrate that the FL explanation differs from CL with the client anomaly percentage.",

keywords = "6G, Privacy, Explainable AI, Federated Learning, Centralized Learning, Security",

author = "Yasintha Rumesh and Senevirathna, \{Thulitha Theekshana\} and Pawani Porambage and Madhusanka Liyanage and Mika Ylianttila",

note = "Publisher Copyright: {\textcopyright} 2023 IEEE.; IEEE International Conference on Communications, ICC 2023 ; Conference date: 28-05-2023 Through 01-06-2023",

year = "2023",

doi = "10.1109/ICC45041.2023.10278845",

language = "English",

pages = "4853--4859",

editor = "Michele Zorzi and Meixia Tao and Walid Saad",

booktitle = "ICC 2023 - IEEE International Conference on Communications",

publisher = "IEEE Institute of Electrical and Electronic Engineers",

address = "United States",

}

Rumesh, Y, Senevirathna, TT, Porambage, P, Liyanage, M & Ylianttila, M 2023, Comprehensive Analysis Over Centralized and Federated Learning-Based Anomaly Detection in Networks with Explainable AI (XAI). in M Zorzi, M Tao & W Saad (eds), ICC 2023 - IEEE International Conference on Communications. IEEE Institute of Electrical and Electronic Engineers, pp. 4853-4859, IEEE International Conference on Communications, ICC 2023, Rome, Italy, 28/05/23. https://doi.org/10.1109/ICC45041.2023.10278845

Comprehensive Analysis Over Centralized and Federated Learning-Based Anomaly Detection in Networks with Explainable AI (XAI). / Rumesh, Yasintha; Senevirathna, Thulitha Theekshana; Porambage, Pawani et al.
ICC 2023 - IEEE International Conference on Communications. ed. / Michele Zorzi; Meixia Tao; Walid Saad. IEEE Institute of Electrical and Electronic Engineers, 2023. p. 4853-4859.

Research output: Chapter in Book/Report/Conference proceeding › Conference article in proceedings › Scientific › peer-review

TY - GEN

T1 - Comprehensive Analysis Over Centralized and Federated Learning-Based Anomaly Detection in Networks with Explainable AI (XAI)

AU - Rumesh, Yasintha

AU - Senevirathna, Thulitha Theekshana

AU - Porambage, Pawani

AU - Liyanage, Madhusanka

AU - Ylianttila, Mika

PY - 2023

Y1 - 2023

N2 - Many forms of machine learning (ML) and artificial intelligence (AI) techniques are adopted in communication networks to perform all optimizations, security management, and decision-making tasks. Instead of using conventional blackbox models, the tendency is to use explainable ML models that provide transparency and accountability. Moreover, Federate Learning (FL) type ML models are becoming more popular than the typical Centralized Learning (CL) models due to the distributed nature of the networks and security privacy concerns. Therefore, it is very timely to research how to find the explainability using Explainable AI (XAI) in different ML models. This paper comprehensively analyzes using XAI in CL and FL-based anomaly detection in networks. We use a deep neural network as the black-box model with two data sets, UNSW-NB15 and NSLKDD, and SHapley Additive exPlanations (SHAP) as the XAI model. We demonstrate that the FL explanation differs from CL with the client anomaly percentage.

AB - Many forms of machine learning (ML) and artificial intelligence (AI) techniques are adopted in communication networks to perform all optimizations, security management, and decision-making tasks. Instead of using conventional blackbox models, the tendency is to use explainable ML models that provide transparency and accountability. Moreover, Federate Learning (FL) type ML models are becoming more popular than the typical Centralized Learning (CL) models due to the distributed nature of the networks and security privacy concerns. Therefore, it is very timely to research how to find the explainability using Explainable AI (XAI) in different ML models. This paper comprehensively analyzes using XAI in CL and FL-based anomaly detection in networks. We use a deep neural network as the black-box model with two data sets, UNSW-NB15 and NSLKDD, and SHapley Additive exPlanations (SHAP) as the XAI model. We demonstrate that the FL explanation differs from CL with the client anomaly percentage.

KW - 6G

KW - Privacy

KW - Explainable AI

KW - Federated Learning

KW - Centralized Learning

KW - Security

UR - https://www.scopus.com/pages/publications/85178296427

U2 - 10.1109/ICC45041.2023.10278845

DO - 10.1109/ICC45041.2023.10278845

M3 - Conference article in proceedings

SP - 4853

EP - 4859

BT - ICC 2023 - IEEE International Conference on Communications

A2 - Zorzi, Michele

A2 - Tao, Meixia

A2 - Saad, Walid

PB - IEEE Institute of Electrical and Electronic Engineers

T2 - IEEE International Conference on Communications, ICC 2023

Y2 - 28 May 2023 through 1 June 2023

ER -

Rumesh Y, Senevirathna TT, Porambage P, Liyanage M, Ylianttila M. Comprehensive Analysis Over Centralized and Federated Learning-Based Anomaly Detection in Networks with Explainable AI (XAI). In Zorzi M, Tao M, Saad W, editors, ICC 2023 - IEEE International Conference on Communications. IEEE Institute of Electrical and Electronic Engineers. 2023. p. 4853-4859 doi: 10.1109/ICC45041.2023.10278845

Comprehensive Analysis Over Centralized and Federated Learning-Based Anomaly Detection in Networks with Explainable AI (XAI)

Abstract

Conference

Funding

Keywords

Access to Document

Other files and links

Fingerprint

SUNSET-6G : Sustainable Network Security Tech for 6G

Cite this

Comprehensive Analysis Over Centralized and Federated Learning-Based Anomaly Detection in Networks with Explainable AI (XAI)

Abstract

Conference

Funding

Keywords

Access to Document

Other files and links

Fingerprint

Projects

SUNSET-6G : Sustainable Network Security Tech for 6G

Cite this