Adaptasi Model Elastic Weight Consolidation Done Right pada Intrusion Detection System

Rahmani, Aulia Daffa (2026) Adaptasi Model Elastic Weight Consolidation Done Right pada Intrusion Detection System. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5025221205-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (6MB)

Abstract

IDS merupakan sistem yang dirancang untuk mendeteksi dan menganalisis aktivitas mencurigakan yang mencoba masuk ke dalam jaringan atau sistem dengan tujuan mencuri, merusak, atau memanipulasi data. Mekanisme kerja akan meliputi proses identifikasi anomali serta deteksi pola serangan yang tidak wajar, kemudian memberikan peringatan apabila ditemukan perilaku yang menyimpang dari standar operasional. Menggunakan DNN sebagai solusi untuk menangani sistem IDS sudah sering dilakukan, akan tetapi pada dunia nyata DNN kerap tidak bisa menyimpan informasi terdahulu atau biasa disebut catastrophic forgetting. Oleh karena itu, penelitian ini mengusulkan arsitektur baru untuk melakukan deteksi covariate shift untuk perubahan data. Continual learrning akan digunakan untuk mengatasi perubahan data dan catastrophic forgetting sehingga sistem dapat bekerja dengan baik pada dunia nyata. Metode yang diusulkan akan dievaluasi dengan NSL-KDD dan UNSW-NB15 dengan metrik evaluasi recall, precision dan F1-Score.
===============================================================================================================================
An IDS is a system designed to detect and analyze suspicious activity attempting to infiltrate a network or system with the aim of stealing, damaging, or manipulating data. Its working mechanism involves identifying anomalies and detecting unusual attack patterns, then providing warnings if behavior deviates from standard operating procedures is found. Using DNN as a solution to handle IDS systems has been frequently done, however, in the real world, DNNs often cannot retain previous information, commonly known as catastrophic forgetting. Therefore, this study proposes a new architecture for detecting covariate shift for data changes. Continual learning will be used to address data changes and catastrophic forgetting so that the system can perform well in the real world. The proposed method will be evaluated on the NSL-KDD and UNSW-NB15 dataset with evaluation metrics through recall, precision, and F1-Score.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Covariate Shift, Continual Learning, Deep Neural Network,Covariate Shift, Continual Learning, Deep Neural Network
Subjects:	T Technology > T Technology (General)
Divisions:	Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Informatics Engineering > 55201-(S1) Undergraduate Thesis
Depositing User:	Aulia Daffa Rahmani
Date Deposited:	28 Jul 2026 01:53
Last Modified:	28 Jul 2026 01:53
URI:	http://repository.its.ac.id/id/eprint/138160

Available Versions of this Item

• Adaptasi Model Elastic Weight Consolidation Done Right pada Intrusion Detection System. (deposited 27 Jul 2026 08:07)
  • Adaptasi Model Elastic Weight Consolidation Done Right pada Intrusion Detection System. (deposited 28 Jul 2026 01:53) [Currently Displayed]

Actions (login required)

View Item