Risk Accessment of Machine Learning Algorithms on Manipulated Dataset in Power Systems
Osuva_Diaba_Shafie-Khah_Mekkanen_Vartiainen_Elmusrati_2023.pdf - Hyväksytty kirjoittajan käsikirjoitus - 472.51 KB
Pysyvä osoite
Kuvaus
©2023 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works.
The emergence of the communication infrastructure in power systems has increased the variety and sophistication of network assaults. Intrusion Detection Systems’ (IDS) importance has increased in relation to network security. IDS, however, is no longer secure when confronted with adversarial examples, and attackers can boost assault success rates by tricking the IDS. As a result, resilience must be increased. This paper assesses the Decision Tree, Logistic regression, Support Vector Machines (SVM), Naïve Bayes, K-Nearest Neighbours (KNN), and Ensemble’s effectiveness. Using the WUSTL-IIoT-2021 dataset and CIC-IDS2017 dataset, we train the algorithms on the unmanipulated dataset and then train the algorithms on the manipulated dataset. Per the simulation results, the accuracy and prediction speed drop on the manipulated dataset while the training time rises.
Emojulkaisu
2023 International Conference on Future Energy Solutions (FES)
ISBN
979-8-3503-3230-8
ISSN
Aihealue
OKM-julkaisutyyppi
A4 Artikkeli konferenssijulkaisussa