The Development of a Firewall Application Using Deep Learning Methods for Detection of Cyber Attacks on Web Applications

Şengül Bayrak; Alper Karaca

Makale detayları

The Development of a Firewall Application Using Deep Learning Methods for Detection of Cyber Attacks on Web Applications

Yazarlar:

Şengül Bayrak | Alper Karaca

Yayıncı:

Dicle University

Ana Başlıklar:

Derin Öğrenme | Doğal Dil İşleme

Dergi:

Dicle Üniversitesi Mühendislik Fakültesi Mühendislik Dergisi Vol.16 Issue 2, pp.331-344

Belge Tipi:

Makale

Yayın Tarihi:

6/2025

Belge Dili:

Türkçe

Document ID (Org No.):

2001352754

pISSN:

1309-8640

Anahtar kelimeler:

Özet:

 The security of web applications is critical for protecting user data and taking precautions against cyber-attacks. Firewall applications are used to protect Internet users sensitive data, take precautions against cyber-attacks and improve the user experience. Although traditional firewall approaches try to detect attacks based on certain rules and patterns, they may be insufficient against evolving and complex attack types. In this paper, we propose a deep learning based approach to detect cyber attacks on web applications. The proposed method is developed using the FWAF dataset and goes through stages that are data preprocessing, feature extraction, and data scaling. Six different deep learning models are evaluated to determine the method with the highest accuracy and reliability. The results obtained played an important role in the selection of the optimum method by comparing the model performances. The selected model was put live on the Jetson Nano platform to test its real-time operability and its performance was evaluated. Experimental results show that deep learning-based approaches offer higher accuracy rates compared to traditional methods and can adapt to new types of attacks. This study makes an important contribution to the field of cyber security by providing an effective and dynamic security solution for detecting attacks against web applications.

 The security of web applications is critical for protecting user data and taking precautions against cyber-attacks. Firewall applications are used to protect Internet users sensitive data, take precautions against cyber-attacks and improve the user experience. Although traditional firewall approaches try to detect attacks based on certain rules and patterns, they may be insufficient against evolving and complex attack types. In this paper, we propose a deep learning based approach to detect cyber attacks on web applications. The proposed method is developed using the FWAF dataset and goes through stages that are data preprocessing, feature extraction, and data scaling. Six different deep learning models are evaluated to determine the method with the highest accuracy and reliability. The results obtained played an important role in the selection of the optimum method by comparing the model performances. The selected model was put live on the Jetson Nano platform to test its real-time operability and its performance was evaluated. Experimental results show that deep learning-based approaches offer higher accuracy rates compared to traditional methods and can adapt to new types of attacks. This study makes an important contribution to the field of cyber security by providing an effective and dynamic security solution for detecting attacks against web applications.

 The security of web applications is critical for protecting user data and taking precautions against cyber-attacks. Firewall applications are used to protect Internet users sensitive data, take precautions against cyber-attacks and improve the user experience. Although traditional firewall approaches try to detect attacks based on certain rules and patterns, they may be insufficient against evolving and complex attack types. In this paper, we propose a deep learning based approach to detect cyber attacks on web applications. The proposed method is developed using the FWAF dataset and goes through stages that are data preprocessing, feature extraction, and data scaling. Six different deep learning models are evaluated to determine the method with the highest accuracy and reliability. The results obtained played an important role in the selection of the optimum method by comparing the model performances. The selected model was put live on the Jetson Nano platform to test its real-time operability and its performance was evaluated. Experimental results show that deep learning-based approaches offer higher accuracy rates compared to traditional methods and can adapt to new types of attacks. This study makes an important contribution to the field of cyber security by providing an effective and dynamic security solution for detecting attacks against web applications.

 Web uygulamalarının güvenliği, kullanıcı verilerinin korunması ve siber saldırılara karşı önlem alınması açısından kritik bir öneme sahiptir. İnternet kullanıcılarının hassas verilerini koruma, siber saldırılara karşı önlem alma ve kullanıcı deneyimini iyileştirme gibi amaçlarla güvenlik duvarı uygulamaları kullanılmaktadır. Geleneksel güvenlik duvarı yaklaşımları, belirli kurallar ve kalıplara dayanarak saldırıları tespit etmeye çalışsa da gelişen ve karmaşık hale gelen saldırı türlerine karşı yetersiz kalabilmektedir. Bu çalışmada, web uygulamalarına yönelik siber saldırıları tespit etmek için derin öğrenme tabanlı bir yaklaşım önerilmektedir. Önerilen yöntem, FWAF veri seti kullanılarak geliştirilmiş ve veri ön işleme, özellik çıkarımı ve veri ölçekleme aşamalarından geçirilmiştir. Altı farklı derin öğrenme modeli değerlendirilerek en yüksek doğruluk ve güvenilirliği sağlayan yöntem belirlenmiştir. Elde edilen sonuçlar, model performanslarını karşılaştırmalı olarak inceleyerek optimum yöntemin seçiminde önemli bir rol oynamıştır. Seçilen model, gerçek zamanlı çalışabilirliğini test etmek amacıyla Jetson Nano platformu üzerinde canlıya alınarak performansı değerlendirilmiştir. Deneysel sonuçlar, derin öğrenme tabanlı yaklaşımların geleneksel yöntemlere kıyasla daha yüksek doğruluk oranları sunduğunu ve yeni saldırı türlerine adapte olabildiğini göstermektedir. Bu çalışma, web uygulamalarına yönelik saldırıların tespitinde etkili ve dinamik bir güvenlik çözümü sunarak, siber güvenlik alanında önemli bir katkı sağlamaktadır.

Makale detayları

The Development of a Firewall Application Using Deep Learning Methods for Detection of Cyber Attacks on Web Applications

Yazarlar:

Şengül Bayrak | Alper Karaca

Yayıncı:

Dicle University

Ana Başlıklar:

Derin Öğrenme | Doğal Dil İşleme

Dergi:

Dicle Üniversitesi Mühendislik Fakültesi Mühendislik Dergisi Vol.16 Issue 2, pp.331-344

Belge Tipi:

Makale

Yayın Tarihi:

6/2025

Belge Dili:

Türkçe

Document ID (Org No.):

2001352754

pISSN:

1309-8640

machine learning

deep learning

firewall

intrusion detection

cybersecurity

web applicationmakine öğrenmesi

derin öğrenme

güvenlik duvarı

saldırı tespiti

siber güvenlik

web uygulaması

Sayı

Makale detayları

Yazarlar: Şengül Bayrak | Alper Karaca

Yayıncı: Dicle University

Ana Başlıklar: Derin Öğrenme | Doğal Dil İşleme

Dergi: Dicle Üniversitesi Mühendislik Fakültesi Mühendislik Dergisi Vol.16 Issue 2, pp.331-344

Belge Tipi: Makale

Yayın Tarihi: 6/2025

Belge Dili: Türkçe

Document ID (Org No.): 2001352754

pISSN: 1309-8640

Sayı

Yazarlar:

Şengül Bayrak | Alper Karaca

Yayıncı:

Dicle University

Ana Başlıklar:

Derin Öğrenme | Doğal Dil İşleme

Dergi:

Dicle Üniversitesi Mühendislik Fakültesi Mühendislik Dergisi Vol.16 Issue 2, pp.331-344

Belge Tipi:

Makale

Yayın Tarihi:

6/2025

Belge Dili:

Türkçe

Document ID (Org No.):

2001352754

pISSN:

1309-8640