• Atmosphere
• /
• v.31 no.1
• /
• pp.73-83
• /
• 2021

Numerical weather prediction (NWP) models play an essential role in predicting weather factors, but using them is challenging due to various factors. To overcome the difficulties of NWP models, deep learning models have been deployed in weather forecasting by several recent studies. This study adapts long short-term memory (LSTM), which demonstrates remarkable performance in time-series prediction. The combination of LSTM model input of meteorological features and activation functions have a significant impact on the performance therefore, the results from 5 combinations of input features and 4 activation functions are analyzed in 9 Automated Surface Observing System (ASOS) stations corresponding to cities/islands/mountains. The optimized LSTM model produces better performance within eight forecast hours than Local Data Assimilation and Prediction System (LDAPS) operated by Korean meteorological administration. Therefore, this study illustrates that this LSTM model can be usefully applied to very short-term weather forecasting, and further studies about CNN-LSTM model with 2-D spatial convolution neural network (CNN) coupled in LSTM are required for improvement.

• Journal of KIISE:Information Networking
• /
• v.37 no.3
• /
• pp.167-174
• /
• 2010

Recently the distributed topology control algorithm for constructing the Recrystallization Topology in the unstructured peer-to-peer network was proposed. In this paper, we prove that such a hierarchical topology is a scale-free power-law network. We present a model of a construction process of the distributed protocol, and analyze it based on a mean-field approximation and the continuum theory, so that we show that the constructed Recrystallization Topology is a scale-free network. In the proposed model, all nodes are born with some initial attractiveness and the system incorporates the rewiring of some links at every time step. Some old links are removed with the anti-preferential probability, and some new links are added with preferential probability. In other words, according to the distributed algorithm, each node makes connections to the more-preferential nodes having higher hit-ratio than other nodes, while it disconnects the anti-preferential nodes having lesser hit-ratio. This gives a realistic description of the local processes forming the recrystallization topology in unstructured peer-to-peer network. We calculate analytically the degree distribution. The analytic result indicates that the constructed network is a scale-free network, of which the scaling exponent is 3.

• Investigative Magnetic Resonance Imaging
• /
• v.15 no.2
• /
• pp.110-122
• /
• 2011

Purpose : The purpose of this study is to establish the method generating human brain anatomical connectivity from Korean children and evaluating the network topological properties using small-world network analysis. Materials and Methods : Using diffusion tensor images (DTI) and parcellation maps of structural MRIs acquired from twelve healthy Korean children, we generated a brain structural connectivity matrix for individual. We applied one sample t-test to the connectivity maps to derive a representative anatomical connectivity for the group. By spatially normalizing the white matter bundles of participants into a template standard space, we obtained the anatomical brain network model. Network properties including clustering coefficient, characteristic path length, and global/local efficiency were also calculated. Results : We found that the structural connectivity of Korean children group preserves the small-world properties. The anatomical connectivity map obtained in this study showed that children group had higher intra-hemispheric connectivity than inter-hemispheric connectivity. We also observed that the neural connectivity of the group is high between brain stem and motorsensory areas. Conclusion : We suggested a method to examine the anatomical brain network of Korean children group. The proposed method can be used to evaluate the efficiency of anatomical brain networks in people with disease.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.8 no.5
• /
• pp.1004-1009
• /
• 2004

Precise time synchronization is a main technology in high-speed communications, parallel and distributed processing systems, Internet information industry and electronic commerce. Synchronized clocks are useful for many leasers. Often a distributed system is designed to realize some synchronized behavior, especially in real-time processing in factories, aircraft, space vehicles, and military applications. Nowadays, time synchronization has been compulsory thing as distributed processing and network operations are generalized. A network time server obtains, keeps accurate and precise time by synchronizing its local clock to standard reference time source and distributes time information through standard time synchronization protocol. This paper describes design issues and implementation of a network time server for time synchronization especially based on a clock model. The system uses GPS (Global Positioning System) as a standard reference time source and offers UTC (universal Time coordinated) through NTP (Network Time protocol). Implementation result and performance analysis are also presented.

• Journal of the Korean earth science society
• /
• v.26 no.3
• /
• pp.268-275
• /
• 2005

One of the nonlinear statistical modelling, neural network method was applied to predict the Sea Surface Temperature Anomalies (SSTA) in the Nino regions, which represent El Nino indices. The data used as inputs in the training step of neural network model were the first seven empirical orthogonal functions in the tropical Pacific $(120^{\circ}\;E,\;20^{\circ}\;S-20^{\circ}\;N)$ obtained from the NCEP/NCAR reanalysis data. The period of 1951 to 1993 was adopted for the training of neural network model, and the period 1994 to 2003 for the forecasting validation. Forecasting results suggested that neural network models were resonable for SSTA forecasting until 9-month lead time. They also predicted greatly the development and decay of strong E1 Nino occurred in 1997-1998 years. Especially, Nino3 region appeared to be the best forecast region, while the forecast skills rapidly decreased since 9-month lead time. However, in the Nino1+2 region where they are relatively low by the influence of local effects, they did not decrease even after 9-month lead time.

• Journal of the Economic Geographical Society of Korea
• /
• v.16 no.4
• /
• pp.567-584
• /
• 2013

This study aims to explore an evolutionary path of the IT service industry in Fairfax County using the Cluster Adaptive Cycle model in economic geography. The analysis is based on detailed historical and industrial information obtained through a variety of data sources including local archival materials, economic census, and interviews. This study also performs a shift-share analysis during the period of 1990 to 2011. Using the adaptive cycle model, the local IT service industry is indicated by a trajectory of constant cluster mutation. The evolution of the local IT service industry has been closely related to federal government policy due to the regional specificity of the National Capital Region and the proximity of the Department of Defense. Although the economic downturn of the late 2000s, the local IT service industry has been notable resilience and adapted to a changing market and technological environment. This constant mutation of the local industry is resulted from not only high resilience which is based on the large government procurement market, the reinforcement of adaptive capacity of the local firms and the network of economic agents such as firm and supporting institutions, but also high flexibility of the knowledge-based service industry to a changing business environment.

• Journal of the Korea Society of Computer and Information
• /
• v.16 no.6
• /
• pp.99-109
• /
• 2011

Ubiquitous computing technology makes an impact on the appearance of u-learning and presents an advanced direction of futuristic school education. In ubiquitous learning environments, various embedded computational devices will be pervasive and interoperable across the network for supporting the learning, so users may utilize these devices anytime anywhere. An important next step for ubiquitous learning is the introduction of context-aware learning service that employing knowledge and reasoning to understand the local context and share this information in support of intelligent learning services. However, the existing studies on design and application of ontology context model to support context-aware service in actual school environments are incomplete state. This paper, therefore, suggests a scheme of constructing ubiquitous learning space for existing school network by introducing USN to support context-aware ubiquitous learning services. This paper, also, designs an ontology based context model for ubiquitous school environments which describes context information through OWL. To determine the suitability of proposed ubiquitous learning space and ontology context model, we implement some of context-aware learning services in the ubiquitous learning environments.

• The Journal of the Acoustical Society of Korea
• /
• v.42 no.3
• /
• pp.214-220
• /
• 2023

In this paper, we propose a method to improve texture recognition performance from haptic acceleration signals representing the texture characteristics of object surface materials by using a Conformer model that combines the advantages of a convolutional neural network and a transformer. In the proposed method, three-axis acceleration signals generated by impact sound and vibration are combined into one-dimensional acceleration data while a person contacts the surface of the object materials using a tool such as a stylus , and the logarithmic Mel-spectrogram is extracted from the haptic acceleration signal similar to the audio signal. Then, Conformer is applied to the extracted the logarithmic Mel-spectrogram to learn main local and global frequency features in recognizing the texture of various object materials. Experiments on the Lehrstuhl für Medientechnik (LMT) haptic texture dataset consisting of 60 materials to evaluate the performance of the proposed model showed that the proposed method can effectively recognize the texture of the object surface material better than the existing methods.

• Korea Trade Review
• /
• v.42 no.1
• /
• pp.47-67
• /
• 2017

This paper analyzes industries of the Daegu-Gyeongbuk province of the Republic of Korea using the product space model that was developed mid 2000s on the basis of a network analysis framework. The study examines the structure of the product space for the 421 export items at HS 4-digit level after grouping them into 4 categories; classics, emerging champions, marginals and the disappearing products. The results suggest a significant structural transformation in the product space for the local industries and the scope and magnitude of such transformation was rather large. We were also able to confirm that the structural transformation of the product space differs from industry to industry and from group to group. While the approach used is not without shortcomings, our findings also indicate that information extracted from the world merchandise trade by way of product space indicators can be quite useful in identifying structural transformation of industries.

• Journal of Intelligence and Information Systems
• /
• v.27 no.3
• /
• pp.57-73
• /
• 2021

Maintenance and prevention of failure through anomaly detection of ICT infrastructure is becoming important. System monitoring data is multidimensional time series data. When we deal with multidimensional time series data, we have difficulty in considering both characteristics of multidimensional data and characteristics of time series data. When dealing with multidimensional data, correlation between variables should be considered. Existing methods such as probability and linear base, distance base, etc. are degraded due to limitations called the curse of dimensions. In addition, time series data is preprocessed by applying sliding window technique and time series decomposition for self-correlation analysis. These techniques are the cause of increasing the dimension of data, so it is necessary to supplement them. The anomaly detection field is an old research field, and statistical methods and regression analysis were used in the early days. Currently, there are active studies to apply machine learning and artificial neural network technology to this field. Statistically based methods are difficult to apply when data is non-homogeneous, and do not detect local outliers well. The regression analysis method compares the predictive value and the actual value after learning the regression formula based on the parametric statistics and it detects abnormality. Anomaly detection using regression analysis has the disadvantage that the performance is lowered when the model is not solid and the noise or outliers of the data are included. There is a restriction that learning data with noise or outliers should be used. The autoencoder using artificial neural networks is learned to output as similar as possible to input data. It has many advantages compared to existing probability and linear model, cluster analysis, and map learning. It can be applied to data that does not satisfy probability distribution or linear assumption. In addition, it is possible to learn non-mapping without label data for teaching. However, there is a limitation of local outlier identification of multidimensional data in anomaly detection, and there is a problem that the dimension of data is greatly increased due to the characteristics of time series data. In this study, we propose a CMAE (Conditional Multimodal Autoencoder) that enhances the performance of anomaly detection by considering local outliers and time series characteristics. First, we applied Multimodal Autoencoder (MAE) to improve the limitations of local outlier identification of multidimensional data. Multimodals are commonly used to learn different types of inputs, such as voice and image. The different modal shares the bottleneck effect of Autoencoder and it learns correlation. In addition, CAE (Conditional Autoencoder) was used to learn the characteristics of time series data effectively without increasing the dimension of data. In general, conditional input mainly uses category variables, but in this study, time was used as a condition to learn periodicity. The CMAE model proposed in this paper was verified by comparing with the Unimodal Autoencoder (UAE) and Multi-modal Autoencoder (MAE). The restoration performance of Autoencoder for 41 variables was confirmed in the proposed model and the comparison model. The restoration performance is different by variables, and the restoration is normally well operated because the loss value is small for Memory, Disk, and Network modals in all three Autoencoder models. The process modal did not show a significant difference in all three models, and the CPU modal showed excellent performance in CMAE. ROC curve was prepared for the evaluation of anomaly detection performance in the proposed model and the comparison model, and AUC, accuracy, precision, recall, and F1-score were compared. In all indicators, the performance was shown in the order of CMAE, MAE, and AE. Especially, the reproduction rate was 0.9828 for CMAE, which can be confirmed to detect almost most of the abnormalities. The accuracy of the model was also improved and 87.12%, and the F1-score was 0.8883, which is considered to be suitable for anomaly detection. In practical aspect, the proposed model has an additional advantage in addition to performance improvement. The use of techniques such as time series decomposition and sliding windows has the disadvantage of managing unnecessary procedures; and their dimensional increase can cause a decrease in the computational speed in inference.The proposed model has characteristics that are easy to apply to practical tasks such as inference speed and model management.