Bui Thu Lam, Le Quy Don Technical University
Deputy Editor-In-Chief
Tran Xuan Nam, Le Quy Don Technical University
Advisory Member
Pham The Long, Le Quy Don Technical University
Nguyen Binh, Posts and Telecommunications Institute of Technology
Dinh The Cuong, Directorate of Information Technology
Luong Chi Mai, Vietnam Academy of Science and Technology
Huynh Quyet Thang, Hanoi University of Science and Technology
Nguyen Huu Thanh, Hanoi University of Science and Technology
Vu Duc Thi, Vietnam National University, Hanoi
Nguyen Thanh Thuy, Vietnam National University, Hanoi
Dao Thanh Tinh, Le Quy Don Technical University
Scientific Secretary
Nguyen Van Giang, Le Quy Don Technical University
Table: Issue 7
ID Paper
[Pdf link]
Mask: New methods to enhance detection of financial fraud.
Nghiem Thi Toan, Nghiem Thi Lich, Bui Duong Hung, Dang Xuan Tho
Nowadays, financial fraud is increasingly popular and causes serious consequences. Therefore, detecting and preventing financial fraud has attracted great attention from researchers. The problem of financial fraud detection can be solved with the support of data mining techniques, such as classification is one of supervised learning methods that is applied most commonly. However, in financial data, the number of samples defined fraud is much fewer than the valid samples, which implies more difficulty of the classification problem. Some well-known methods of solving this problem such as SMOTE, Borderline-SMOTE, and SPY have achieved positive results, but in some cases they cannot improve or sometimes reduce classification performance. In this paper, we propose a new method, MASK, to change the label of a majority class samples based on the density distribution in the minority class samples. The experimental results on international standard datasets such as UCSD-FICO (Data mining Contest 2009), German Credit, Australian Credit, and Yeast (from UCI) also showed that the new method is effective and improves the accuracy of classification of financial data comparing to ROS, RUS, SMOTE, Borderline- SMOTE, and SPY.
Application of neural networks to intrusion detection system using system calls frequency –based with ADFA-LD.
Nguyen Viet Hung.
In this paper, a method for preprocessing the ADFA Linux dataset (ADFA-LD) dataset using frequency-based and PCA method will be introduced. After that, the neural networks algorithm is used to detect intrusion. Results from experiments show that, at the same accuracy rate, our method has the false alarm rate lower than other methods.
Intrusion detection approach using tree adjoining grammar guided genetic programming.
Vu Van Canh, Hoang Tuan Hao, Nguyen Van Quan.
In recent years, network security issues have become urgent and significant impact on the performance of modern computer networks. Network intrusion detection/prevention system has been the topic of many studies researchers worldwide to improve the security of a network. However, the intrusion detection systems are not high effective for new attacks, or variants of known attacks. Machine learning approaches applied in intrusion detection have overcome restrictions on and increasingly shown the superiority in detecting new attacks with many different methods. In this paper, we use genetic programming technique (GP) and Tree Adjoining Grammar Guided Genetic Programming (TAG3P) on artificial datasets from [25]. Based on experimental results and comparisons, we found that GP and TAG3P are more effective in detecting attacks than previous measures.
An advanced probabilistic method using a Bayesian Network Model to evaluate uncertainty in software project scheduling.
Nguyen Ngoc Tuan, Tran Trung Hieu, Huynh Quyet Thang
Software Risk Management has become a vital part of Software Project Management since software development involves uncertainty (or risk factors that might have bad impacts on the project). In fact, all the phases of the software development life cycle (SDLC) are potential sources of uncertainty since they have to deal with hardware, software, technology, people, cost, and processes. To lead a software project to success, it is required to model and assess uncertainty since the early phases of the project. Current state-of-the-art scheduling techniques based on the assumption that every task, activity or phase of the project is carried out exactly as it is planned, which almost never happens in real-life projects. Recent researches on risk management focus on the relationships between uncertainty and the outcomes of a project. This research examines a model and a probabilistic tool CKDY using Bayesian Belief Networks to evaluate risk factors in software project scheduling.
Enhancement of Cooperative Spectrum Sensing Employing Genetic Algorithm and Noise Power Estimation.
Hoang Manh Kha, Nguyen Viet Tuyen, Nguyen Hai Duong, Vo Kim.
In cognitive radio networks (CRN), spectrum sensing is a key functionality to enchance the spectrum efficiency. Principal factors influencing the detection performance of the system in soft-decision fusion based cooperative spectrum sensing are weight coefficients vector. This paper proposes to use Expectation-Maximization algorithm to estimate noise power in case of missing data combined with genetic algorithm to optimize weight vectors by maximizing the probability of detection. The simulation results demonstrate that the proposed method outperforms the traditional methods in the sense of the performance of energy detection based spectrum sensing for CRN.
Asynchronous wireless replay network with phase feedback.
Tran The Nghiep.
The wireless relay network with different propagation delays from relay nodes to destination node is called the asynchronous wireless relay network. Due to the effect of different propagation delays among relay nodes that leads to inter-symbol interferences (ISI) in received signal at the destination node and reduces significantly the system performance. In this paper, we propose a new wireless relay network with phase feedback where received signals at relay nodes are multiplied by feedback phase from destination node before being forwarded to destination node. This allows the transmission from relay nodes to the destination to perform in only one time slot and reducing the number of ISI components to a minimum (it means only one ISI component in the received signal). As results, the proposed phase feedback scheme enables robust against asynchronous, improves significantly array gain, and outperforms the previous distributed close loop extended orthogonal space-time block code.
Energy management system with low power loss and reliable data transmisstion based on power line communication.
Vo Minh Huan.
The monitoring and calculating electricity consumption by manual method is still very popular in Vietnam. This method presents many disadvantages such as: time consuming, uncontrolling power consumption of the load and household consumption, being difficult to detect the behavior of electricity fraud. So, we decided to research and build a power consumption management system that could solve these problems. The system uses Power Line Communication technology that allows direct data transmission over power lines existing. At the same time, we have also developed a reliable data transmission protocol to collect data as accurately as possible. With this system, power monitoring becomes very easy, when the customer's electricity parameters are posted to the website, the supervisor monitors the customer's electricity number via the internet. It allows the manager to shut down the power of each customer when a problem occurs or not pay without affecting other customers.