🌐 Within the framework of the 24th International Symposium on Communications and Information Technologies (ISCIT 2025),
held at Vietnam National University, Hanoi, from October 16–18, 2025, two student teams from the AI Institute – VNU University of Engineering and Technology (UET, VNU) had their papers accepted and presented at this prestigious IEEE-indexed international conference.
🌟 Team 1 – Cyberattack Detection on Blockchain Smart Contracts and Transactions
Student: Lê Vũ Hiếu (K68A-AI1)
Co-authors / Supervisors: MSc. Đỗ Hải Sơn (Curtin University, Australia), Dr. Trần Việt Khoa
The team focuses on developing a method for cyberattack detection in smart contracts and transactions on blockchain systems. Their approach applies an NLP-based preprocessing technique to convert smart contract bytecode into opcode sequences, which are subsequently vectorized using TF-IDF. These features are combined with transaction-level attributes, such as gas and input, to form a unified, image-like representation.
This representation is fed into a Vision Transformer (ViT) model enhanced with a Convolutional Embedding Block, allowing the model to capture complex attack patterns and significantly improve detection performance.
🌟 Team 2 – α-OPIT: Robust Online Subspace Tracking for Streaming Data
Students:
– Tạ Giang Thùy Loan (K68A-AI1)
– Nguyễn Hoàng Lan (K68-EC2)
– Nguyễn Thị Ngọc Lan (K68A-AI2)
Supervisors / Co-authors:
Dr. Lê Trung Thành (UET-VNU),
MSc. Đỗ Hải Sơn (Curtin University, Australia),
Dr. Trần Thị Thúy Quỳnh (UET-VNU),
Prof. Nguyễn Linh Trung (UET-VNU),
Prof. Karim Abed-Meraim (University of Orléans, France)
The research proposes α-OPIT, an improved version of the Online Power Iteration via Thresholding (OPIT) algorithm, by integrating α-divergence to enhance robustness against non-Gaussian noise and outliers commonly found in streaming data.
The method enables stable and efficient estimation of evolving sparse subspaces while remaining computationally optimal—demonstrating strong potential in signal processing, online learning, and real-time data analysis applications.
🎉 Congratulations!
The outstanding achievements of students from the AI Institute at an IEEE-indexed international conference reflect their research capability, creativity, and continuous effort. Congratulations to all participating students for this remarkable accomplishment. We hope their success will inspire more young researchers at UET to pursue scientific discovery and innovation in the years ahead.
— AI Institute, VNU-UET
