Deteksi Serangan Brute Force dengan Support Vector Machine (SVM) pada Sistem Otentikasi

Satria, Fajar (2025) Deteksi Serangan Brute Force dengan Support Vector Machine (SVM) pada Sistem Otentikasi. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 05111940000083-Undergraduate_Theses.pdf Restricted to Repository staff only Download (7MB) | Request a copy

Abstract

Keamanan sistem otentikasi merupakan faktor krusial dalam melindungi akun pengguna dari akses tidak sah. Salah satu ancaman paling umum terhadap sistem otentikasi adalah serangan brute force, di mana penyerang mencoba berbagai kombinasi nama pengguna dan kata sandi secara berulang hingga menemukan kredensial yang valid. Serangan ini dapat menyebabkan pengambilalihan akun, pencurian data, atau bahkan akses tidak sah ke sistem yang lebih luas. Metode keamanan tradisional, seperti pembatasan jumlah percobaan login, penggunaan CAPTCHA, serta pendekatan berbasis aturan, sering kali kurang efektif dalam menghadapi serangan otomatis yang semakin canggih. Untuk meningkatkan deteksi serangan brute force, diperlukan pendekatan berbasis pembelajaran mesin yang mampu mengenali pola serangan secara lebih adaptif. Salah satu algoritma yang banyak digunakan dalam klasifikasi keamanan siber adalah Support Vector Machine (SVM), yang memiliki keunggulan dalam memisahkan data berdasarkan pola yang kompleks. Tugas akhir ini menerapkan algoritma Support Vector Machine (SVM) untuk mendeteksi serangan brute force pada sistem otentikasi dengan memanfaatkan fitur-fitur yang direkayasa dari log login. Berdasarkan hasil evaluasi, model SVM dengan konfigurasi Nonlinear (RBF kernel) tanpa pembobotan kelas menghasilkan performa terbaik dalam membedakan aktivitas login normal dan serangan. Selain itu, pendekatan berbasis pembelajaran mesin ini terbukti lebih akurat dan responsif dibandingkan metode konvensional, khususnya metode berbasis aturan, dalam mendeteksi serangan secara lebih dini dan tepat sasaran.
=====================================================================================================================================
Authentication system security is crucial in protecting user accounts from unauthorized access. One of the most common threats to authentication systems is brute-force attacks, in which attackers repeatedly try various username and password combinations until they find valid credentials. These attacks can lead to account takeovers, data theft, or even unauthorized access to broader systems. Traditional security methods, such as limiting login attempts, using CAPTCHAs, and rule-based approaches, are often ineffective against increasingly sophisticated automated attacks. To improve brute-force attack detection, a machine learning-based approach capable of more adaptively recognizing attack patterns is needed. One algorithm widely used in cybersecurity classification is the Support Vector Machine (SVM), which excels at separating data based on complex patterns. This final project applies the Support Vector Machine (SVM) algorithm to detect brute-force attacks on authentication systems by utilizing engineered features from login logs. Based on the evaluation results, the SVM model with a nonlinear configuration (RBF kernel) without class weighting performed best in distinguishing between normal login activity and attacks. In addition, this machine learning-based approach has proven to be more accurate and responsive than conventional methods, especially rule-based methods, in detecting attacks earlier and more precisely.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Support Vector Machine, Brute Force, Sistem Otentikasi, Pembelajaran Mesin, Deteksi Serangan Siber
Subjects:	T Technology > T Technology (General) > T57.8 Nonlinear programming. Support vector machine. Wavelets. Hidden Markov models.
Divisions:	Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Informatics Engineering > 55201-(S1) Undergraduate Thesis
Depositing User:	Fajar Satria
Date Deposited:	29 Jul 2025 04:15
Last Modified:	29 Jul 2025 04:15
URI:	http://repository.its.ac.id/id/eprint/122590

Actions (login required)

View Item