• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.16 no.3
• /
• pp.813-829
• /
• 2022

In multi-view subspace clustering, how to integrate the complementary information between perspectives to construct a unified representation is a critical problem. In the existing works, the unified representation is usually constructed in the original data space. However, when the data representation in each view is very diverse, the unified representation derived directly in the original data domain may lead to a huge information loss. To address this issue, different to the existing works, inspired by the latest revelation that the data across all perspectives have a very similar or close spectral block structure, we try to construct the unified representation in the spectral embedding domain. In this way, the complementary information across all perspectives can be fused into a unified representation with little information loss, since the spectral block structure from all views shares high consistency. In addition, to capture the global structure of data on each view with high accuracy and robustness both, we propose a novel low-rank approximation via the tight lower bound on the rank function. Finally, experimental results prove that, the proposed method has the effectiveness and robustness at the same time, compared with the state-of-art approaches.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.15 no.11
• /
• pp.3913-3934
• /
• 2021

Real-time prediction of Web service of quality (QoS) provides more convenience for web services in cloud environment, but real-time QoS prediction faces severe challenges, especially under the cold-start situation. Existing literatures of real-time QoS predicting ignore that the QoS of a user/service is related to the QoS of other users/services. For example, users/services belonging to the same group of category will have similar QoS values. All of the methods ignore the group relationship because of the complexity of the model. Based on this, we propose a real-time Matrix Factorization based Clustering model (MFC), which uses category information as a new regularization term of the loss function. Specifically, in order to meet the real-time characteristic of the real-time prediction model, and to minimize the complexity of the model, we first map the QoS values of a large number of users/services to a lower-dimensional space by the PCA method, and then use the K-means algorithm calculates user/service category information, and use the average result to obtain a stable final clustering result. Extensive experiments on real-word datasets demonstrate that MFC outperforms other state-of-the-art prediction algorithms.

• The Journal of the Acoustical Society of Korea
• /
• v.21 no.2
• /
• pp.199-210
• /
• 2002

In this paper, we carried out the study on speech recognition using the KM-Net topology design algorithm based on decision tree state-clustering to improve the performance of acoustic models in speech recognition. The Korean has many allophonic and grammatical rules compared to other languages, so we investigate the allophonic variations, which defined the Korean phonetics, and construct the phoneme question set for phonetic decision tree. The basic idea of the HM-Net topology design algorithm is that it has the basic structure of SSS (Successive State Splitting) algorithm and split again the states of the context-dependent acoustic models pre-constructed. That is, it have generated. the phonetic decision tree using the phoneme question sets each the state of models, and have iteratively trained the state sequence of the context-dependent acoustic models using the PDT-SSS (Phonetic Decision Tree-based SSS) algorithm. To verify the effectiveness of the above algorithm we carried out the speech recognition experiments for 452 words of center for Korean language Engineering (KLE452) and 200 sentences of air flight reservation task (YNU200). Experimental results show that the recognition accuracy has progressively improved according to the number of states variations after perform the splitting of states in the phoneme, word and continuous speech recognition experiments respectively. Through the experiments, we have got the average 71.5%, 99.2% of the phoneme, word recognition accuracy when the state number is 2,000, respectively and the average 91.6% of the continuous speech recognition accuracy when the state number is 800. Also we haute carried out the word recognition experiments using the HTK (HMM Too1kit) which is performed the state tying, compared to share the parameters of the HM-Net topology design algorithm. In word recognition experiments, the HM-Net topology design algorithm has an average of 4.0% higher recognition accuracy than the context-dependent acoustic models generated by the HTK implying the effectiveness of it.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.10 no.6
• /
• pp.2669-2688
• /
• 2016

An efficient and effective unsupervised single moving object detection framework is presented in this paper. Given the sparsely labelled trajectory points, we adopt a coarse-to-fine strategy to detect and segment the foreground from the background. The superpixel level coarse segmentation reduces the complexity of subsequent processing, and the pixel level refinement improves the segmentation accuracy. A distance measurement is devised in the coarse segmentation stage to measure the similarities between generated superpixels, which can then be used for clustering. Moreover, a Quadmap is introduced to facilitate the refinement in the fine segmentation stage. According to the experiments, our algorithm is effective and efficient, and favorable results can be achieved compared with state-of-the-art methods.

• Journal of Korea Multimedia Society
• /
• v.10 no.12
• /
• pp.1645-1654
• /
• 2007

In this paper, we study the problem of load balancing routing in clustered-based wireless mesh network in order to enhance the overall network throughput. We first address the problems of cluster allocation in wireless mesh network to achieve load-balancing state. Due to the complexity of the problem, we proposed a simplified algorithm using gradient load-balancing model. This method searches for a localized optimal solution of cluster allocation instead of solving the optimal solution for overall network. To support for load-balancing algorithm and reduce complexity of topology control, we also introduce limited broadcasting between two clusters. This mechanism maintain shortest path between two nodes in adjacent clusters while minimizing the topology broadcasting complexity. The simulation experiments demonstrate that our proposed model achieve performance improvement in terms of network throughput in comparison with other clustering methods.

• Journal of the Optical Society of Korea
• /
• v.14 no.4
• /
• pp.298-304
• /
• 2010

The pair-induced quenching (PIQ) effect in a highly doped bismuth oxide-based erbium-doped fiber amplifier (EDFA) was theoretically and experimentally investigated. In the theoretical investigation, the bismuth oxide-based EDFA was modeled as a 6-level amplifier system that incorporated clustering-induced concentration quenching, cooperative up-conversion, pump excited state absorption (ESA), and signal ESA. The relative number of paired ions in a highly doped bismuth oxide EDF was estimated to be ~6.02%, determined by a comparison between the theoretical and the experimentally measured gain values. The impacts of the PIQ on the gain and the noise figure were also investigated.

• The Bulletin of The Korean Astronomical Society
• /
• v.35 no.1
• /
• pp.82-82
• /
• 2010

We measure the topology of the galaxy distribution using the Seventh Data Release of the Sloan Digital Sky Survey (SDSS DR7), examining the dependence of galaxy clustering topology on galaxy properties. The observational results are used to test galaxy formation models. A volume-limited sample defined by Mr<-20.19 enables us to measure the genus curve with amplitude of G=378 at 6h-1Mpc smoothing scale, with 4.8% uncertainty including all systematics and cosmic variance. The clustering topology over the smoothing length interval from 6 to 10h-1Mpc reveals a mild scale-dependence for the shift and void abundance (A_V) parameters of the genus curve. We find strong bias in the topology of galaxy clustering with respect to the predicted topology of the matter distribution, which is also scale-dependent. The luminosity dependence of galaxy clustering topology discovered by Park et al. (2005) is confirmed: the distribution of relatively brighter galaxies shows a greater prevalence of isolated clusters and more percolated voids. We find that galaxy clustering topology depends also on morphology and color. Even though early (late)-type galaxies show topology similar to that of red (blue) galaxies, the morphology dependence of topology is not identical to the color dependence. In particular, the void abundance parameter A_V depends on morphology more strongly than on color. We test five galaxy assignment schemes applied to cosmological N-body simulations to generate mock galaxies: the Halo-Galaxy one-to-one Correspondence (HGC) model, the Halo Occupation Distribution (HOD) model, and three implementations of Semi-Analytic Models (SAMs). None of the models reproduces all aspects of the observed clustering topology; the deviations vary from one model to another but include statistically significant discrepancies in the abundance of isolated voids or isolated clusters and the amplitude and overall shift of the genus curve. SAM predictions of the topology color-dependence are usually correct in sign but incorrect in magnitude.

• KIPS Transactions on Software and Data Engineering
• /
• v.8 no.1
• /
• pp.45-50
• /
• 2019

With the gradual advancement of IT, passive work processes are automated and the overall quality of life has greatly improved. This is made possible by the formation of an organic topology between a wide variety of real-life smart devices. To serve these diverse smart devices, businesses or users are using the cloud. The services in the cloud are divided into Infrastructure as a Service (IaaS), Platform as a Service (PaaS) and Software as a Service (SaaS). SaaS runs on PaaS, and PaaS runs on IaaS. Since IaaS is the basis of all services, an algorithm is required to operate virtualization resources efficiently. Among them, desktop resource virtualization is used for resource high availability of unused state time of existing desktop PC. Clustering of hierarchical structures is important for high availability of these resources. In addition, it is very important to select a suitable algorithm because many clustering algorithms are mainly used depending on the distribution ratio and environment of the desktop PC. If various attempts are made to find an algorithm suitable for desktop resource virtualization in an operating environment, a great deal of power, time, and manpower will be incurred. Therefore, this paper proposes a resource clustering simulator for cluster selection of desktop virtualization. This provides a clustering simulation to properly select clustering algorithms and apply elements in different environments of desktop PCs.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.10 no.2
• /
• pp.504-521
• /
• 2016

Recently, energy harvesting technology has been integrated into wireless sensor networks to ameliorate the nodes' energy limitation problem. In theory, the wireless sensor node equipped with an energy harvesting module can work permanently until hardware failures happen. However, due to the change of power supply, the traditional hierarchical network routing protocol can not be effectively adopted in energy harvesting wireless sensor networks. In this paper, we improve the Low-Energy Adaptive Clustering Hierarchy (LEACH) protocol to make it suitable for the energy harvesting wireless sensor networks. Specifically, the cluster heads are selected according to the estimation of nodes' harvested energy and consumed energy. Preference is given to the nodes with high harvested energy while taking the energy consumption rate into account. The utilization of harvested energy is mathematically formulated as a max-min optimization problem which maximizes the minimum energy conservation of each node. We have proved that maximizing the minimum energy conservation is an NP-hard problem theoretically. Thus, a polynomial time algorithm has been proposed to derive the near-optimal performance. Extensive simulation results show that our proposed routing scheme outperforms previous works in terms of energy conservation and balanced distribution.

• Journal of Microbiology and Biotechnology
• /
• v.18 no.2
• /
• pp.194-198
• /
• 2008

A newly designed Cytophaga-Flavobacteria-specific 16S rRNA gene primer pair was employed to investigate the CF community structure in the Bering Sea, revealing a previously unknown and unexpected high CF diversity in this high latitude cold sea. In total, 56 clones were sequenced and 50 unique CF 16 rRNA gene fragments were obtained, clustering into 16 CF subgroups, including nine cosmopolitan subgroups, five psychrophilic subgroups, and two putatively autochthonous subgroups. The majority of sequences (82%) were closely related to uncultured CF species and could not be classified into known CF genera, indicating the presence of a large number of so-far uncultivated CF species in the Bering Sea.