• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.17 no.10
• /
• pp.2252-2258
• /
• 2013

In this paper, parallel orthogonal matching pursuit (POMP) is proposed to supplement the simultaneous orthogonal matching pursuit (S-OMP) which has been widely used as a greedy algorithm for sparse signal recovery for multiple measurement vector (MMV) problem. The process of POMP is simple but effective: (1) multiple indexes maximally correlated with the observation vector are chosen at the first iteration, (2) the conventional S-OMP process is carried out in parallel for each selected index, (3) the index set which yields the minimum residual is selected for reconstructing the original sparse signal. Empirical simulations show that POMP for MMV outperforms than the conventional S-OMP both in terms of exact recovery ratio (ERR) and mean-squared error (MSE).

• KIPS Transactions on Computer and Communication Systems
• /
• v.4 no.9
• /
• pp.289-296
• /
• 2015

The explosive growth of information has been difficult for users to get an appropriate information in time. The various ways of new services to solve problems has been provided. As customized service is being magnified, the personalized recommendation system has been important issue. Collaborative filtering system in the recommendation system is widely used, and it is the most successful process in the recommendation system. As the recommendation is based on customers' profile, there can be sparsity and cold-start problems. In this paper, we propose personalized movie recommendation system using collaborative filtering techniques and context-based techniques. The context-based technique is the recommendation method that considers user's environment in term of time, emotion and location, and it can reflect user's preferences depending on the various environments. In order to utilize the context-based technique, this paper uses the human emotion, and uses movie reviews which are effective way to identify subjective individual information. In this paper, this proposed method shows outperforming existing collaborative filtering methods.

• KIISE Transactions on Computing Practices
• /
• v.24 no.2
• /
• pp.93-98
• /
• 2018

Recommendation systems are used to provide items of interests for users to maximize a company's profit. Matrix factorization is frequently used by recommendation systems, based on an incomplete user-item rating matrix. However, as the number of items and users increase, it becomes difficult to make accurate recommendations due to the sparsity of data. To overcome this drawback, the use of text data related to items was recently suggested for matrix factorization algorithms. Furthermore, a word-level convolutional neural network was shown to be effective in the process of extracting the word-level features from the text data among these kinds of matrix factorization algorithms. However, it involves a large number of parameters to learn in the word-level convolutional neural network. Thus, we propose a matrix factorization algorithm which utilizes a character-level convolutional neural network with which to extract the character-level features from the text data. We also conducted a performance study with real-life datasets to show the effectiveness of the proposed matrix factorization algorithm.

• Journal of Digital Contents Society
• /
• v.16 no.4
• /
• pp.605-613
• /
• 2015

This paper addresses the problem to make sparse the projection matrix in pattern recognition method. Recently, the size of computer program is often restricted in embedded systems. It is very often that developed programs include some constant data. For example, many pattern recognition programs use the projection matrix for dimension reduction. To improve the recognition performance, very high dimensional feature vectors are often extracted. In this case, the projection matrix can be very big. Recently, RSR(roated sparse regression) method[1] was proposed. This method has been proved one of the best algorithm that obtains the sparse matrix. We propose three methods to improve the RSR; outlier removal, sampling and elastic net RSR(E-RSR) in which the penalty term in RSR optimization function is replaced by that of the elastic net regression. The experimental results show that the proposed methods are very effective and improve the sparsity rate dramatically without sacrificing the recognition rate compared to the original RSR method.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.41 no.8
• /
• pp.917-920
• /
• 2016

SpSF algorithm is direction-of-arrival estimation algorithm based on sparse representation of incident signlas. Cost function to be optimized for DOA estimation is multi-dimensional nonlinear function, which is hard to handle for optimization. After some manipulation, the problem can be cast into convex optimiztion problem. Convex optimization problem tuns out to be constrained optimization problem, where the parameter in the constraint has to be determined. The solution of the convex optimization problem is dependent on the specific parameter value in the constraint. In this paper, we propose a rule-of-thumb for determining the parameter value in the constraint. Based on the fact that the noise in the array elements is complex Gaussian distributed with zero mean, the average of the Frobenius norm of the matrix in the constraint can be rigorously derived. The parameter in the constrint is set to be two times the average of the Frobenius norm of the matrix in the constraint. It is shown that the SpSF algorithm actually works with the parameter value set by the method proposed in this paper.

• Journal of the Institute of Convergence Signal Processing
• /
• v.9 no.4
• /
• pp.295-303
• /
• 2008

In general, transform domain adaptive filters show faster convergence speed than the time domain adaptive filters, but the amount of calculation increases dramatically as the filter order increases. This problem can be solved by making use of the subband structure in transform domain adaptive filters. In this paper, to increase the convergence speed on the generalized subband decomposition FIR adaptive filters, a structure of the adaptive filter with subfilter of dyadic sparsity factor in wavelet transform domain is designed. And, in this adaptive filter, the equivalent input in transform domain is derived and, by using the input, the convergence properties for the LMS algorithm is analyzed and evaluated. By using this sub band adaptive filter, the inverse system modeling and the periodic noise canceller were designed, and, by computer simulation, the convergence speeds of the systems on LMS algorithm were compared with that of the subband adaptive filter using DFT(discrete Fourier transform).

• Proceedings of the Korean Society of Computational and Applied Mathematics Conference
• /
• 2003.09a
• /
• pp.1-1
• /
• 2003

In this talk, a reduced-order modeling methodology based on centroidal Voronoi tessellations (CVT's)is introduced. CVT's are special Voronoi tessellations for which the generators of the Voronoi diagram are also the centers of mass (means) of the corresponding Voronoi cells. The discrete data sets, CVT's are closely related to the h-means clustering techniques. Even with the use of good mesh generators, discretization schemes, and solution algorithms, the computational simulation of complex, turbulent, or chaotic systems still remains a formidable endeavor. For example, typical finite element codes may require many thousands of degrees of freedom for the accurate simulation of fluid flows. The situation is even worse for optimization problems for which multiple solutions of the complex state system are usually required or in feedback control problems for which real-time solutions of the complex state system are needed. There hava been many studies devoted to the development, testing, and use of reduced-order models for complex systems such as unsteady fluid flows. The types of reduced-ordered models that we study are those attempt to determine accurate approximate solutions of a complex system using very few degrees of freedom. To do so, such models have to use basis functions that are in some way intimately connected to the problem being approximated. Once a very low-dimensional reduced basis has been determined, one can employ it to solve the complex system by applying, e.g., a Galerkin method. In general, reduced bases are globally supported so that the discrete systems are dense; however, if the reduced basis is of very low dimension, one does not care about the lack of sparsity in the discrete system. A discussion of reduced-ordering modeling for complex systems such as fluid flows is given to provide a context for the application of reduced-order bases. Then, detailed descriptions of CVT-based reduced-order bases and how they can be constructed of complex systems are given. Subsequently, some concrete incompressible flow examples are used to illustrate the construction and use of CVT-based reduced-order bases. The CVT-based reduced-order modeling methodology is shown to be effective for these examples and is also shown to be inexpensive to apply compared to other reduced-order methods.

• Journal of KIISE:Software and Applications
• /
• v.31 no.8
• /
• pp.1062-1072
• /
• 2004

Collaborative filtering systems based on ｛user-document｝ matrix are effective in recommending web documents to user. But they have a shortcoming of decreasing the accuracy of recommendations by the first rater problem and the sparsity. This paper proposes the automatic preference rating method that generates user profile to solve the shortcoming. The profile in this paper is content-based collaborative user profile. The content-based collaborative user profile is generated by combining a content-based user profile with a collaborative user profile by mutual information method. Collaborative user profile is based on ｛user-document｝ matrix in collaborative filtering system, thus, content-based user profile is generated by relevance feedback in content-based filtering systems. After normalizing combined content-based collaborative user profiles, it automatically rates user preference by reflecting normalized profile in ｛user-document｝matrix of collaborative filtering systems. We evaluated our method on a large database of user ratings for web document and it was certified that was more efficient than existent methods.