• Journal of KIISE
• /
• v.44 no.11
• /
• pp.1231-1235
• /
• 2017

Given the network structure in online social network, it is important to determine a way to distinguish spam accounts from the network features. In online social network, the service provider attempts to detect social spamming to maintain their service quality. However the spammer group changes their strategies to avoid being detected. Even though the spammer attempts to act as legitimate users, certain distinguishable structural features are not easily changed. In this paper, we investigate a way to generate meaningful network structure features, and suggest spammer detection method using recursive structural features. From a result of real-world dataset experiment, we found that the proposed algorithm could improve the classification performance by about 8%.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2019.06a
• /
• pp.98-101
• /
• 2019

At present, deep convolutional neural networks have made a very important contribution in single-image super-resolution. Through the learning of the neural networks, the features of input images are transformed and combined to establish a nonlinear mapping of low-resolution images to high-resolution images. Some previous methods are difficult to train and take up a lot of memory. In this paper, we proposed a simple and compact deep recursive residual network learning the features for single image super resolution. Global residual learning and local residual learning are used to reduce the problems of training deep neural networks. And the recursive structure controls the number of parameters to save memory. Experimental results show that the proposed method improved image qualities that occur in previous methods.

• Structural Engineering and Mechanics
• /
• v.80 no.6
• /
• pp.711-726
• /
• 2021

This paper presents a multiple parameters-based recursive methodology for the reliability quantification of civil structures subjected to meteorological disasters. Recognizing the challenge associated with characterizing at a single stroke all the meteorological disasters that may hit a structure during its service life, the proposed methodology by contrast features a multiparameter recursive mechanism to describe the meteorological demand of the structure. The benefit of the arrangements is that the essentially inevitable deviation of the practically observed meteorological data from those in the existing model can be mitigated in an adaptive manner. In particular, the implications of potential climate change to the relevant reliability of civil structures are allowed for. The application of the formulated methodology of recursive reliability quantification is illustrated by first considering the reliability quantification of a linear shear frame against simulated strong wind loads. A parametric study is engaged in this application to examine the effect of some hyperparameters in the configured hierarchical model. Further, the application is extended to a nonlinear hysteretic shear frame involving some field-observed cyclone data, and the incompleteness of the relevant structural diagnosis data that may arise in reality is taken into account. Also investigated is another application scenario where the reliability of a building envelope is assessed under hailstone impacts, and the emphasis is to demonstrate the recursive incorporation of newly obtained meteorological data.

• Journal of Information Processing Systems
• /
• v.17 no.2
• /
• pp.337-351
• /
• 2021

Automatically recognizing facial expressions in video sequences is a challenging task because there is little direct correlation between facial features and subjective emotions in video. To overcome the problem, a video facial expression recognition method using spatiotemporal recurrent neural network and feature fusion is proposed. Firstly, the video is preprocessed. Then, the double-layer cascade structure is used to detect a face in a video image. In addition, two deep convolutional neural networks are used to extract the time-domain and airspace facial features in the video. The spatial convolutional neural network is used to extract the spatial information features from each frame of the static expression images in the video. The temporal convolutional neural network is used to extract the dynamic information features from the optical flow information from multiple frames of expression images in the video. A multiplication fusion is performed with the spatiotemporal features learned by the two deep convolutional neural networks. Finally, the fused features are input to the support vector machine to realize the facial expression classification task. The experimental results on cNTERFACE, RML, and AFEW6.0 datasets show that the recognition rates obtained by the proposed method are as high as 88.67%, 70.32%, and 63.84%, respectively. Comparative experiments show that the proposed method obtains higher recognition accuracy than other recently reported methods.

• Journal of the Korean Institute of Telematics and Electronics B
• /
• v.30B no.11
• /
• pp.17-26
• /
• 1993

In this paper, a VlSI architecture for Reed-Solomon (RS) decoder based on the Berlekamp algorithm is proposed. The proposed decoder provided both erasure and error correcting capability. In order to reduc the chip area, we reformulate the Berlekamp algorithm. The proposed algorithm possesses a recursive structure so that the number of cells for computing the errata locator polynomial can be reduced. Moreover, in our approach, only one finite field multiplication per clock cycle is required for implementation, provided an improvement in the decoding speed, and the overall architecture features parallel and pipelined structure, making a real time decoding possible. From the performance evaluation, it is concluded that the proposed VLSI architecture is more efficient in terms of VLSI implementation than the rcursive architecture based on the Euclid algorithm.

• Asia pacific journal of information systems
• /
• v.12 no.2
• /
• pp.99-118
• /
• 2002

In this paper, we propose a visual relational database query language, VRQL, by modifying and extending the recently proposed $VOQL^*$. Like $VOQL^*$, VRQL, based on ven Diagram and graph, naturally reflects the structure of schemas in queries and has recursive formal semantics. However, VRQL has relationally complete expressiveness, while $VOQL^*$ is only a conjunctive query language. In the logical definition part of VRQL, which is the relational version of $VOQL^*$, most features of $VOQL^*$ are retained, and the semantics of queries are based on the tuple relational calculus. In the procedural definition part of VRQL, by introducing the concept of VRQL view and set operations, the expressiveness of VRQL is increased to the level equivalent to that of the relational algebra. Due to the introduction of VRQL views, existing queries or temporary queries used in the process of creating queries can be represented with views, so that complex queries may be represented more conveniently. Set operations, used with VRQL views, enable us to represent various queries, beyond the expressiveness of conjunctive query languages.

• Korean Journal of Computational Design and Engineering
• /
• v.10 no.5
• /
• pp.303-313
• /
• 2005

This paper proposes a generalized NURBS model, called Volumetric NURBS or VNURBS for representing volumetric objects with multiple attributes embedded in multidimensional space. This model provides a mathematical framework for modeling complex structure of heterogeneous objects and analyzing inside of objects to discover features that are directly inaccessible, for deeper understanding of complex field configurations. The defining procedure of VNURBS, which explains two directional extensions of NURBS, shows VNURBS is a generalized volume function not depending on the domain and its range dimensionality. And the recursive a1gorithm for VNURBS derivatives is described as a computational basis for efficient and robust volume modeling. In addition, the specialized versions of VNURBS demonstrate that VNURBS is applicable to various applications such as geometric modeling, volume rendering, and physical field modeling.

• Communications for Statistical Applications and Methods
• /
• v.19 no.2
• /
• pp.293-301
• /
• 2012

Quantile regression proposed by Koenker and Bassett (1978) is a statistical technique that estimates conditional quantiles. The advantage of using quantile regression is the robustness in response to large outliers compared to ordinary least squares(OLS) regression. A regression tree approach has been applied to OLS problems to fit flexible models. Loh (2002) proposed the GUIDE algorithm that has a negligible selection bias and relatively low computational cost. Quantile regression can be regarded as an analogue of OLS, therefore it can also be applied to GUIDE regression tree method. Chaudhuri and Loh (2002) proposed a nonparametric quantile regression method that blends key features of piecewise polynomial quantile regression and tree-structured regression based on adaptive recursive partitioning. Lee and Lee (2006) investigated wage determinants in the Korean labor market using the Korean Labor and Income Panel Study(KLIPS). Following Lee and Lee, we fit three kinds of quantile regression tree models to KLIPS data with respect to the quantiles, 0.05, 0.2, 0.5, 0.8, and 0.95. Among the three models, multiple linear piecewise quantile regression model forms the shortest tree structure, while the piecewise constant quantile regression model has a deeper tree structure with more terminal nodes in general. Age, gender, marriage status, and education seem to be the determinants of the wage level throughout the quantiles; in addition, education experience appears as the important determinant of the wage level in the highly paid group.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.29 no.11C
• /
• pp.1518-1526
• /
• 2004

In this paper, an algorithm using wavelet transform for detecting a cut that is a radical scene transition point, and fade and dissolve that are gradual scene transition points is proposed. The conventional methods using wavelet transform for this purpose is using features in both spatial and frequency domain. But in the proposed algorithm, the color space of an input image is converted to YUV and then luminance component Y is transformed in frequency domain using 2-level lifting. Then, the histogram of only low frequency subband that may contain some spatial domain features is compared with the previous one. Edges obtained from other higher bands can be divided into global, semi-global and local regions and the histogram of each edge region is compared. The experimental results show the performance improvement of about 17% in recall and 18% in precision and also show a good performance in fade and dissolve detection.

• Journal of Broadcast Engineering
• /
• v.26 no.4
• /
• pp.429-440
• /
• 2021

With the recent development of deep convolutional neural network learning, deep learning techniques applied to single image super-resolution are showing good results. One of the existing deep learning-based super-resolution techniques is RDN(Residual Dense Network), in which the initial feature information is transmitted to the last layer using residual dense blocks, and subsequent layers are restored using input information of previous layers. However, if all hierarchical features are connected and learned and a large number of residual dense blocks are stacked, despite good performance, a large number of parameters and huge computational load are needed, so it takes a lot of time to learn a network and a slow processing speed, and it is not applicable to a mobile system. In this paper, we use the residual dense structure, which is a continuous memory structure that reuses previous information, and the residual dense channel attention block using the channel attention method that determines the importance according to the feature map of the image. We propose a method that can increase the depth to obtain a large receptive field and maintain a concise model at the same time. As a result of the experiment, the proposed network obtained PSNR as low as 0.205dB on average at 4× magnification compared to RDN, but about 1.8 times faster processing speed, about 10 times less number of parameters and about 1.74 times less computation.