• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.6 no.10
• /
• pp.2650-2662
• /
• 2012

Due to the semantic gap issue, the performance of automatic image annotation is still far from satisfactory. Active learning approaches provide a possible solution to cope with this problem by selecting most effective samples to ask users to label for training. One of the key research points in active learning is how to select the most effective samples. In this paper, we propose a novel active learning approach based on sparse graph. Comparing with the existing active learning approaches, the proposed method selects the samples based on two criteria: uncertainty and representativeness. The representativeness indicates the contribution of a sample's label propagating to the other samples, while the existing approaches did not take the representativeness into consideration. Extensive experiments show that bringing the representativeness criterion into the sample selection process can significantly improve the active learning effectiveness.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.10 no.12
• /
• pp.5401-5421
• /
• 2016

Both low-density parity-check (LDPC) codes and the multiple access technique of spatially coupling data transmission (SCDT) can be expressed in bipartite graphs. To improve the performance of iterative demodulation and decoding for SCDT, a novel joint sparse graph (JSG) with SCDT and LDPC codes is constructed. Based on the JSG, an approach for iterative joint demodulation and decoding by belief propagation (BP) is presented as an exploration of the flooding schedule, and based on BP, density evolution equations are derived to analyze the performance of the iterative receiver. To accelerate the convergence speed and reduce the complexity of joint demodulation and decoding, a novel serial schedule is proposed. Numerical results show that the joint demodulation and decoding for SCDT based on JSG can significantly improve the system's performance, while roughly half of the iterations can be saved by using the proposed serial schedule.

• Communications of the Korean Institute of Information Scientists and Engineers
• /
• v.5 no.2
• /
• pp.85-89
• /
• 1987

The large sparse matrix problems arise in many applications areas, such as structural analysis, network analysis. In dealing with such sparse systems proper preprogramming techniques such as permuting rows and columns simultaneously, will be needed in order to reduce the number of arithmetic operations and storage spaces.

• Journal of Institute of Control, Robotics and Systems
• /
• v.18 no.12
• /
• pp.1096-1100
• /
• 2012

Most of previous genetic algorithms for solving graph problems have used a vertex-based encoding. We proposed an edge encoding based new genetic algorithm using a spanning tree. Contrary to general edge-based encoding, a spanning tree-based encoding represents only feasible partitions. As a target problem, we adopted the MAX CUT problem, which is well known as a representative NP-hard problem, and examined the performance of the proposed genetic algorithm. The experiments on benchmark graphs are executed and compared with vertex-based encoding. Performance improvements of the spanning tree-based encoding on sparse graphs was observed.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.8 no.9
• /
• pp.3211-3230
• /
• 2014

Recently, sparse coding has achieved remarkable success in image representation tasks. In practice, the performance of clustering can be significantly improved if limited label information is incorporated into sparse coding. To this end, in this paper, a novel semi-supervised algorithm, called constrained sparse concept coding (CSCC), is proposed for image representation. CSCC considers limited label information into graph embedding as additional hard constraints, and hence obtains embedding results that are consistent with label information and manifold structure information of the original data. Therefore, CSCC can provide a sparse representation which explicitly utilizes the prior knowledge of the data to improve the discriminative power in clustering. Besides, a kernelized version of our proposed CSCC, namely kernel constrained sparse concept coding (KCSCC), is developed to deal with nonlinear data, which leads to more effective clustering performance. The experimental evaluations on the MNIST, PIE and Yale image sets show the effectiveness of our proposed algorithms.

• Journal of Multimedia Information System
• /
• v.3 no.3
• /
• pp.53-61
• /
• 2016

This research proposes the index optimization as a classification task and application of the graph based KNN. We need the index optimization as an important task for maximizing the information retrieval performance. And we try to solve the problems in encoding words into numerical vectors, such as huge dimensionality and sparse distribution, by encoding them into graphs as the alternative representations to numerical vectors. In this research, the index optimization is viewed as a classification task, the similarity measure between graphs is defined, and the KNN is modified into the graph based version based on the similarity measure, and it is applied to the index optimization task. As the benefits from this research, by modifying the KNN so, we expect the improvement of classification performance, more graphical representations of words which is inherent in graphs, the ability to trace more easily results from classifying words. In this research, we will validate empirically the proposed version in optimizing index on the two text collections: NewsPage.com and 20NewsGroups.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.9 no.7
• /
• pp.2568-2584
• /
• 2015

Spectral clustering has become one of the most popular clustering approaches in recent years. Similarity graph constructed on the data is one of the key factors that influence the performance of spectral clustering. However, the similarity graphs constructed by existing methods usually contain some unreliable edges. To construct reliable similarity graph for spectral clustering, an efficient method based on Markov random walk (MRW) is proposed in this paper. In the proposed method, theMRW model is defined on the raw k-NN graph and the neighbors of each sample are determined by the probability of the MRW. Since the high order transition probabilities carry complex relationships among data, the neighbors in the graph determined by our proposed method are more reliable than those of the existing methods. Experiments are performed on the synthetic and real-world datasets for performance evaluation and comparison. The results show that the graph obtained by our proposed method reflects the structure of the data better than those of the state-of-the-art methods and can effectively improve the performance of spectral clustering.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2018.05a
• /
• pp.41-43
• /
• 2018

소설 데이터는 인터넷 상의 수 많은 개인과 개인의 상호 작용에 의하여 연결되어 있으며, 이러한 데이터를 분석하여, 분석 대상에 내재하고 있는 구조와 특성을 파악하는 일은 중요하다. 특히, 개인 추천을 위해서는 개별 데이터들의 관계 그래프를 활용하여 빠르고 정확하게 추천 값을 도출하는 것이 효율적이다. 하지만, 기존 추천 기법으로는 신규 사용자와 아이템이 끊임없이 등장하는 상황을 즉각적으로 반영하기가 어렵고, 또한 많은 결측값을 포함하는 sparse 한 데이터일 경우에는 추천 시스템의 연산 공간과 시간에 많은 제약이 있다. 이에 본 논문에서는 Spark GraphX 를 활용한 개인 추천 시스템을 설계 및 개발하였으며, 이를 통하여 사용자와 아이템간에 내재하는 복합 요인이 반영된 그래프 기반 추천을 실행하여, 개인 추천 결과의 우수성을 확인하였다.

• 한국태양에너지학회:학술대회논문집
• /
• 2012.03a
• /
• pp.419-424
• /
• 2012

This paper discusses the state-of-the-art techniques in real-time state estimation for the Smart Microgrids. The most popular method used in traditional power system state estimation is a Weighted Least Square(WLS) algorithm which is based on Maximum Likelihood(ML) estimation under the assumption of static system state being a set of deterministic variables. In this paper, we present a survey of dynamic state estimation techniques for Smart Microgrids based on Belief Propagation (BP) when the system state is a set of stochastic variables. The measurements are often too sparse to fulfill the system observability in the distribution network of microgrids. The BP algorithm calculates posterior distributions of the state variables for real-time sparse measurements. Smart Microgrids are modeled as a factor graph suitable for characterizing the linear correlations among the state variables. The state estimator performs the BP algorithm on the factor graph based the stochastic model. The factor graph model can integrate new models for solar and wind correlation. It provides the Smart Microgrids with a way of integrating the distributed renewable energy generation. Our study on Smart Microgrid state estimation can be extended to the estimation of unbalanced three phase distribution systems as well as the optimal placement of smart meters.

• Journal of KIISE:Computing Practices and Letters
• /
• v.16 no.4
• /
• pp.432-436
• /
• 2010

Finding communities from network data including social networks can be done by clustering the nodes of the network as densely interconnected groups, where keeping interconnection between groups sparse. To exploit a clustering algorithm for community detection task, we need a well-defined similarity measure between network nodes. In this paper, we propose a new similarity measure named "Common Neighborhood Sub-graph density" and combine the similarity with affinity propagation, which is a recently devised clustering algorithm.