Skip to content
  • Tiếng Việt
  • English

Congratulations to the two students from the Faculty of Computer Networks and Communications for their scientific paper published at the International Conference on Advanced Technologies for Communications (ATC) 2023.

Paper Title: "IncWAD: An Incremental Learning Approach for Web Attack Detection using MLOps"

Authors:

Le Phi Hung – 21522115 – MMCL2021 – Lead Author

Vu Xuan Khang – 21522202 – MMCL2021 – Co-Author

Supervisors:

Dr. Pham Van Hau

M.Sc. Nghi Hoang Khoa

Abstract:

With the continuous development of the Internet and web applications, they have become an indispensable part of our daily lives. However, this ubiquity has also led to sophisticated and malicious web attacks, necessitating the development of highly effective Web Attack Detection (WAD) systems. Several Machine Learning (ML) based approaches to WAD have been explored and proven to be highly efficient. Furthermore, to ensure optimal classification performance with data containing new labels, these models need frequent retraining. However, retraining models with both old and new data poses significant challenges in terms of computational complexity and storage requirements. Conversely, retraining models solely with new data leads to Catastrophic Forgetting (CF), where previously learned knowledge from the old data is lost. To address these challenges, this research proposes an WAD system based on Deep Neural Networks (DNN) along with Incremental Learning (IL) techniques, named IncWAD. Additionally, we design a model management and deployment cycle (MLOps) to automate the training and updating of machine learning models. Evaluation results on the multi-label dataset SR-BH 2020 under various scenarios demonstrate that IncWAD can accurately classify different types of attacks with an accuracy exceeding 95%. In the future, the team proposes the development of a Federated Web Attack Detection (FWAD) model to leverage large-scale data from participating entities without sharing the data, enhancing efficiency in multi-label data classification.

"We sincerely express our gratitude to Dr. Pham Van Hau – Dean of Faculty of Information Security, for creating an incredibly dynamic academic environment for us students. We thank M.Sc. Nghi Hoang Khoa for his guidance, support, invaluable advice, and for accompanying us throughout this significant period. Additionally, we extend our thanks to Mr. Nguyen Huu Quyen – Researcher at InSecLab, for his valuable input during our research. We appreciate the dedicated teaching, identification of limitations, and insightful suggestions provided by all our mentors, helping us achieve the best possible research outcomes."

The International Conference on Advanced Technologies for Communications (ATC) is an annual international conference on advanced technologies for communications. This conference series, initiated in 2008, is jointly organized by the Radio & Electronics Association of Vietnam (REV) and the IEEE Communications Society (IEEE ComSoc). The primary objectives of this conference series are to provide an international forum for exchanging scientific and technological knowledge among researchers and engineers in Vietnam and worldwide in the fields of electronics, communications, and related areas. In 2023, ATC will be hosted by the University of Information Technology and Communications Vietnam-Korea (VKU) from October 19 to October 21, 2023. The conference will feature prominent invited speakers and high-quality research contributions from leading researchers worldwide.

Detailed Information: https://www.facebook.com/UIT.Fanpage/posts/pfbid02KMVr95AqXuoFshYej3PWCfMzMz9KFJEnepPdzRTPPhNKD9dTnYA81FQYT1FDiywMl

Hai Bang - Communication Collaborator, University of Information Technology

English version: Phan Huy Hoang

Tập tin đính kèm: