Nurfajrin, Lilik Citra (2025) OPTIMASI DETEKSI INTRUSI BERBASIS MULTI-LAYER PERCEPTRON (MLP) MELALUI PENYEIMBANGAN DATA DENGAN SYNTHETIC MINORITY OVERSAMPLING TECHNIQUE (SMOTE). S2 thesis, Universitas Mercu Buana Jakarta - Menteng.
|
Text (COVER)
55421120015-LILIK CITRA NURFAJRIN-01 COVER - Lilik Citra Nurfajrin.pdf Download (855kB) | Preview |
|
![[img]](https://repository.mercubuana.ac.id/style/images/fileicons/text.png) |
Text (BAB I)
55421120015-LILIK CITRA NURFAJRIN-02 BAB 1 - Lilik Citra Nurfajrin.pdf Restricted to Registered users only Download (46kB) |
|
|
Text (BAB II)
55421120015-LILIK CITRA NURFAJRIN-03 BAB 2 - Lilik Citra Nurfajrin.pdf Restricted to Registered users only Download (116kB) |
|
|
Text (BAB III)
55421120015-LILIK CITRA NURFAJRIN-04 BAB 3 - Lilik Citra Nurfajrin.pdf Restricted to Registered users only Download (87kB) |
|
|
Text (BAB IV)
55421120015-LILIK CITRA NURFAJRIN-05 BAB 4 - Lilik Citra Nurfajrin.pdf Restricted to Registered users only Download (456kB) |
|
|
Text (BAB V)
55421120015-LILIK CITRA NURFAJRIN-06 BAB 5 - Lilik Citra Nurfajrin.pdf Restricted to Registered users only Download (39kB) |
|
|
Text (DAFTAR PUSTAKA)
55421120015-LILIK CITRA NURFAJRIN-08 DAFTAR PUSTAKA - Lilik Citra Nurfajrin.pdf Restricted to Registered users only Download (168kB) |
|
|
Text (LAMPIRAN)
55421120015-LILIK CITRA NURFAJRIN-09 LAMPIRAN - Lilik Citra Nurfajrin.pdf Restricted to Registered users only Download (143kB) |
Abstract
Pada era perkembangan teknologi saat ini, keamanan jaringan dan keamanan data menjadi aspek yang sangat penting di berbagai sektor. Kemajuan teknologi internet telah memudahkan kehidupan masyarakat, namun juga meningkatkan potensi ancaman dan serangan terhadap keamanan jaringan. Ruang siber menjadi semakin rentan terhadap serangan siber otomatis yang bersifat berkelanjutan dan dapat berasal dari berbagai sumber. Dalam melatih Intrusion Detection System (IDS), dibutuhkan dataset yang besar dan seimbang. Namun, pada kenyataannya, jumlah data kategori serangan jaringan jauh lebih sedikit dibandingkan dengan trafik normal, sehingga menurunkan tingkat deteksi IDS. Kondisi ketidakseimbangan data ini perlu mendapat perhatian khusus untuk menghasilkan sistem pendeteksi intrusi yang optimal. Penelitian ini membahas upaya mitigasi ketidakseimbangan data pada keamanan jaringan menggunakan metode Synthetic Minority Oversampling Technique (SMOTE). Ketidakseimbangan data, yang kerap terjadi pada dataset keamanan jaringan seperti CIC-IDS 2017, dapat memengaruhi kinerja model deteksi intrusi berbasis pembelajaran mesin. Penelitian ini mengevaluasi dampak ketidakseimbangan data terhadap model Multilayer Perceptron (MLP) serta mengimplementasikan SMOTE untuk menyeimbangkan distribusi data antar kelas. Hasil penelitian menunjukkan bahwa penerapan SMOTE secara signifikan meningkatkan akurasi, sensitivitas, dan presisi model deteksi intrusi, dengan akurasi akhir mencapai 98,08%. Temuan ini menegaskan bahwa penyeimbangan data memiliki peran penting dalam meningkatkan efektivitas model pendeteksian ancaman siber. Dengan demikian, penerapan SMOTE dapat menjadi solusi strategis untuk mengatasi masalah ketidakseimbangan data pada pengembangan sistem keamanan jaringan yang andal. In today’s era of technological advancement, network security and data security have become crucial aspects across various sectors. The development of internet technology has made daily life more convenient, yet it has also increased the potential threats and attacks targeting network security. Cyberspace is becoming increasingly vulnerable to automated and persistent cyberattacks originating from various sources. To train an Intrusion Detection System (IDS), a large and balanced dataset is required. However, in reality, the amount of network attack data is significantly smaller than normal traffic data, which lowers the detection rate of IDS. This data imbalance issue requires serious attention to produce an optimal intrusion detection system. This study addresses data imbalance mitigation in network security using the Synthetic Minority Oversampling Technique (SMOTE). Data imbalance, which frequently occurs in network security datasets such as CIC-IDS 2017, can significantly affect the performance of machine learning–based intrusion detection models. This research evaluates the impact of data imbalance on a Multilayer Perceptron (MLP) model and applies SMOTE to balance the class distribution within the dataset. The results demonstrate that the application of SMOTE significantly improves the accuracy, sensitivity, and precision of the intrusion detection model, achieving a final accuracy of 98.08%. These findings emphasize the importance of data balancing in enhancing the effectiveness of cyber threat detection models. Therefore, SMOTE can serve as a strategic solution to address data imbalance issues in the development of robust network security systems.
| Item Type: | Thesis (S2) |
|---|---|
| NIM/NIDN Creators: | 55421120015 |
| Uncontrolled Keywords: | Ketidakseimbangan data, SMOTE, keamanan jaringan, deteksi intrusi, CIC-IDS 2017, machine learning, deep learning, Multilayer Perceptron (MLP). Data imbalance, SMOTE, network security, intrusion detection, CIC-IDS 2017, machine learning, deep learning, Multilayer Perceptron (MLP). |
| Subjects: | 600 Technology/Teknologi > 620 Engineering and Applied Operations/Ilmu Teknik dan operasi Terapan > 621 Applied Physics/Fisika terapan |
| Divisions: | Pascasarjana > Magister Teknik Elektro |
| Depositing User: | NAIMAH NUR ISLAMIDIYANAH |
| Date Deposited: | 04 Sep 2025 03:52 |
| Last Modified: | 04 Sep 2025 03:52 |
| URI: | http://repository.mercubuana.ac.id/id/eprint/97427 |
Actions (login required)
 |
View Item |
