• 응용통계연구
• /
• 제28권1호
• /
• pp.1-8
• /
• 2015

인터넷 트래픽의 증가로 인하여 네트워크의 보안 문제가 중요한 문제로 대두되고 있다. 그 중에서도 특히 P2P 트래픽의 증가는 모든 서버의 관리자에게는 해결해야할 중요한 문제로 대두되고 있다. 서버에서 네트워크 트래픽을 조사하여 문제가 있는 트래픽을 미리 차단하는 것은 서비스 품질의 향상과 자원의 효율적인 사용 측면에서 바람직하나 오가는 패킷의 내부정보를 조사하는 것은 개인정보보호 차원에서 문제가 있을 수 있으며 시간과 노력이 많이 소요되므로 요즘은 통계적인 기계학습의 방법을 이용하여 이상 트래픽을 찾아내는 연구가 주를 이루고 있다. 본 연구에서는 최근의 기계학습방법 중에서 널리 쓰이는 방법들을 비교 연구하여 그 결과 랜덤포리스트(random forest)라고 불리는 방법의 우수함을 보였다.

• Journal of the Korean Physical Society
• /
• 제73권10호
• /
• pp.1589-1595
• /
• 2018

Similarity index measures the topological proximity of node pairs in a complex network. Numerous similarity indices have been defined and investigated, but the dependency of structure on the performance of similarity indices has not been sufficiently investigated. In this study, we investigated the relationship between the performance of similarity indices and structural properties of a network by employing a two-state random network. A node in a two-state network has binary types that are initially given, and a connection probability is determined from the state of the node pair. The performances of similarity indices are affected by the number of links and the ratio of intra-connections to inter-connections. Similarity indices have different characteristics depending on their type. Local indices perform well in small-size networks and do not depend on whether the structure is intra-dominant or inter-dominant. In contrast, global indices perform better in large-size networks, and some such indices do not perform well in an inter-dominant structure. We also found that link prediction performance and the performance of similarity are correlated in both model networks and empirical networks. This relationship implies that link prediction performance can be used as an approximation for the performance of the similarity index when information about node type is unavailable. This relationship may help to find the appropriate index for given networks.

• International Journal of Internet, Broadcasting and Communication
• /
• 제12권4호
• /
• pp.26-32
• /
• 2020

There are many airline customer evaluation data, but they are insufficient in terms of predicting customer satisfaction in practice. In particular, they are generally insufficient in case of verification of data value and development of a customer satisfaction prediction model based on customer evaluation data. In this paper, airline customer satisfaction analysis is conducted through an experiment of correlation analysis between customer evaluation data provided by Google's Kaggle. The difference in accuracy varied according to the three types, which are the overall variables, the top 4 and top 8 variables with the highest correlation. To build an airline customer satisfaction prediction model, they are applied to three classification algorithms of Random Forest, SVM, DNN and conduct a classification experiment. They are divided into training data and verification data by 7:3. As a result, the DNN model showed the lowest accuracy at 86.4%, while the SVM model at 89% and the Random Forest model at 95.7% showed the highest accuracy and performance.

• 한국지하수토양환경학회:학술대회논문집
• /
• 한국지하수토양환경학회 2000년도 추계학술대회
• /
• pp.69-73
• /
• 2000

We systematically analyze the influence of fracture junction, solute transfer characteristics on transport patterns in discrete, two-dimensional fracture network models. Regular lattices and random fracture networks with power-law length distributions are considered in conjunction with particle tracking methods. Solute transfer probabilities at fracture junctions are determined from analytical considerations and from simple complete mixing and streamline routing models. For regular fracture networks, mixing conditions at fracture junctions are always dominated by either complete mixing or streamline routing end member cases. Moreover bulk transport properties such as the spreading and the dilution of solute are highly sensitive to the mixing rule. However in power-law length networks there is no significant difference in bulk transport properties, as calculated by assuming either of the two extreme mixing rules. This apparent discrepancy between the effects of mixing properties at fracture junctions in regular and random fracture networks is explained by the statistics of the coordination number and of the flow conditions at fracture intersections. We suggest that the influence of mixing rules on bulk solute transport could be important in systematic orthogonal fracture networks but insignificant in random networks.

• 대한전자공학회논문지TC
• /
• 제45권9호
• /
• pp.30-38
• /
• 2008

유선망을 가정하여 설계된 기존의 TCP를 무선망이 혼재된 망에서 그대로 사용하면 무선구간의 랜덤 손실도 유선구간의 혼잡 손실과 동일하게 간주하게 된다. 이러한 잘못된 판단과 조치는 대역폭을 낭비하여 전체 네트워크의 처리량(throughput)을 저하하게 된다. 이러한 문제를 하기 위한 노력들이 많이 있었다. 본 논문에서는 유선망 혼잡과 무선구간 손실을 판별할 개선된 방법을 사용하여 망 상황 진단의 정확도를 높이고, 판단된 각 상황에 대해서 망 처리량을 높일 수 있는 개선된 제어 방법을 제안하였다. 또한, 시뮬레이션을 통해 제안한 알고리즘의 성능을 TCP Westwood, TCP Veno와 비교하여 평가하였다.

• International Journal of Computer Science & Network Security
• /
• 제23권10호
• /
• pp.81-88
• /
• 2023

In the present scenario, enormous amounts of data are produced every second. These data also contain private information from sources including media platforms, the banking sector, finance, healthcare, and criminal histories. Data mining is a method for looking through and analyzing massive volumes of data to find usable information. Preserving personal data during data mining has become difficult, thus privacy-preserving data mining (PPDM) is used to do so. Data perturbation is one of the several tactics used by the PPDM data privacy protection mechanism. In Perturbation, datasets are perturbed in order to preserve personal information. Both data accuracy and data privacy are addressed by it. This paper will explore and compare several perturbation strategies that may be used to protect data privacy. For this experiment, two perturbation techniques based on random projection and principal component analysis were used. These techniques include Improved Random Projection Perturbation (IRPP) and Enhanced Principal Component Analysis based Technique (EPCAT). The Naive Bayes classification algorithm is used for data mining approaches. These methods are employed to assess the precision, run time, and accuracy of the experimental results. The best perturbation method in the Nave-Bayes classification is determined to be a random projection-based technique (IRPP) for both the cardiovascular and hypothyroid datasets.

• Structural Engineering and Mechanics
• /
• 제31권2호
• /
• pp.153-162
• /
• 2009

The reduction of the dynamic response of an offshore structure subjected to wind-generated random ocean waves is of extreme significance in the aspects of serviceability, fatigue life and safety of the structure. In this study, a new neuro-control scheme is applied to the vibration control of a fixed offshore platform under random wave loads to examine the applicability of the proposed method. It is called the Lattice Probabilistic Neural Network (LPNN), as it utilizes lattice pattern of state vectors as the training data of PNN. When control results of the LPNN are compared with those of the NN and PNN, LPNN showed better performance in effectively suppressing the structural responses in a shorter computational time.

• 대한기계학회:학술대회논문집
• /
• 대한기계학회 2004년도 추계학술대회
• /
• pp.208-213
• /
• 2004

Methods for estimation of the effective stress intensity factor range (${\Delta}K_{eff}$) are evaluated for narrow and wide band random loading crack growth test data of 2024-T351 aluminum alloy. Three methods of determining $K_{op}$, visual measurement, ASTM offset compliance method, and the neural network method proposed by Kang and Song, and three methods of estimating ${\Delta}K_{eff}$, conventional, the 2/PI0 and 2/PI methods proposed by Donald and Paris, are compared in a quantitative manner by using the results of fatigue crack growth life prediction under random loading. For all $K_{op}$ determination methods discussed, the 2/PI0 and 2/PI methods of estimating ${\Delta}K_{eff}$ provide better results than conventional method for narrow and wide band random loading data.

• 제어로봇시스템학회:학술대회논문집
• /
• 제어로봇시스템학회 2005년도 ICCAS
• /
• pp.941-946
• /
• 2005

In this paper an active random noise control using adaptive learning rate neural networks is presented. The adaptive learning rate strategy increases the learning rate by a small constant if the current partial derivative of the objective function with respect to the weight and the exponential average of the previous derivatives have the same sign, otherwise the learning rate is decreased by a proportion of its value. The use of an adaptive learning rate attempts to keep the learning step size as large as possible without leading to oscillation. It is expected that a cost function minimize rapidly and training time is decreased. Numerical simulations and experiments of active random noise control with the transfer function of the error path will be performed, to validate the convergence properties of the adaptive learning rate Neural Networks. Control results show that adaptive learning rate Neural Networks control structure can outperform linear controllers and conventional neural network controller for the active random noise control.

• 센서학회지
• /
• 제29권4호
• /
• pp.227-231
• /
• 2020

Silver nanowire random networks are promising candidates for replacing indium tin oxide (ITO) as transparent and conductive electrodes. They can also be used as transparent heating films with self-cleaning and defogging properties. By virtue of the Joule heating effect, silver nanowire random networks can be heated when voltage bias is applied; however, they are unsuitable for long-term use. In this work, we study the Joule heating of silver nanowire random networks embedded in polymers. Silver nanowire random networks embedded in polymers exhibit breakdown under the application of electric current. Their surface morphological changes indicate that nanoparticle formation may be the main cause of this electrical breakdown. Numerical analyses are used to investigate the temperatures of the silver nanowire and substrate.