• The Journal of the Korea Contents Association
• /
• v.15 no.5
• /
• pp.560-569
• /
• 2015

Crowd funding is the method of raising money for a project, companies from a large group of people via the Internet, in return for future products or equity. Kickstarter is the largest and most successful crowdfunding site where creative projects raise reward based funding. Drawing on dataset of 80,267 projects with combined funding over $1.3b from 8.1m people, this paper suggest that backer select project based on their preference on the project, instead profitability of the project. It suggests that well-established platform and big size of network increases the chance of success of the project due to a ripple effect and blockbuster effects. Clear communication about the project's idea and goal is highly correlated with success. Regular communication on the project site, such as by constant progress updates, helps the success of the project. Equity-based crowdfunding is emerging as an innovative means of raising capital for businesses, so it has been receiving a lot of attention and expectation from the government and the market. The findings of this paper and others will help to get some understanding and insight into equity-based crowdfunding. However, Kickstarter differs from equity-based crowdfunding in the goals of the backers. Kickstarter's backers are not investors, they are contributors. To understand equity-based crowdfunding, the subject will need further study.

• Geophysics and Geophysical Exploration
• /
• v.10 no.4
• /
• pp.241-251
• /
• 2007

The observed spectra from Odaesan earthquake were fitted to a point-source spectral model to evaluate the source spectrum and spatial features of the modelling error. The source spectrum was calculated by removing from the observed spectra the path and site dependent responses (Yun, 2007) that were previously revealed through an inversion process applied to a large accumulated spectral dataset. The stress drop parameter of one-corner Brune's ${\omega}^2$ source model fitted to the estimated source spectrum was well predicted by the scaling relation between magnitude and stress drop developed by Yun et al. (2006). In particular, the estimated spectrum was quite comparable to the two-corner source model that was empirically developed for recent moderate earthquakes occurring around the Korean Peninsula, which indicates that Odaesan earthquake is one of typical moderate earthquakes representative of Korean Peninsula. Other features of the observed spectra from Odaesan earthquake were also evaluated based on the commonly treated random error between the observed data and the estimated point-source spectral model. Radiation pattern of the error according to azimuth angle was found to be similar to the theoretical estimate. It was also observed that the spatial distribution of the errors was correlated with the geological map and the $Q_0$ map which are indicatives of seismic boundaries.

• Journal of the Korea Society of Computer and Information
• /
• v.19 no.2
• /
• pp.193-200
• /
• 2014

Due to the advent of ubiquitous computing environment, it is becoming a part of our common life style. And tremendous information is cumulated rapidly. In these trends, it is becoming a very important technology to find out exact information in a large data to present users. Collaborative filtering is the method based on other users' preferences, can not only reflect exact attributes of user but also still has the problem of sparsity and scalability, though it has been practically used to improve these defects. In this paper, we propose clustering method by user's features based on SOM for predicting purchase pattern in u-Commerce. it is necessary for us to make the cluster with similarity by user's features to be able to reflect attributes of the customer information in order to find the items with same propensity in the cluster rapidly. The proposed makes the task of clustering to apply the variable of featured vector for the user's information and RFM factors based on purchase history data. To verify improved performance of proposing system, we make experiments with dataset collected in a cosmetic internet shopping mall.

• Journal of the Korea Society of Computer and Information
• /
• v.16 no.11
• /
• pp.245-253
• /
• 2011

Due to the imprudent spending of the fossil fuels, the environment was contaminated seriously and the exhaustion problems of the fossil fuels loomed large. Therefore people become taking a great interest in alternative energy resources which can solve problems of fossil fuels. The wind power energy is one of the most interested energy in the new and renewable energy. However, the plants of wind power energy and the traditional power plants should be balanced between the power generation and the power consumption. Therefore, we need analysis and prediction to generate power efficiently using wind energy. In this paper, we have performed a research to predict power generation patterns using the wind power data. Prediction approaches of datamining area can be used for building a prediction model. The research steps are as follows: 1) we performed preprocessing to handle the missing values and anomalous data. And we extracted the characteristic vector data. 2) The representative patterns were found by the MIA(Mean Index Adequacy) measure and the SOM(Self-Organizing Feature Map) clustering approach using the normalized dataset. We assigned the class labels to each data. 3) We built a new predicting model about the wind power generation with classification approach. In this experiment, we built a forecasting model to predict wind power generation patterns using the decision tree.

• Journal of Korea Water Resources Association
• /
• v.51 no.9
• /
• pp.747-759
• /
• 2018

Multivariate regional frequency analysis has advantages of regional and multivariate framework as adopting a large number of regional dataset and modeling phenomena that cannot be considered in the univariate frequency analysis. To the best of our knowledge, the multivariate regional frequency analysis has not been employed for hydrological variables in South Korea. Applicability of the multivariate regional frequency analysis should be investigated for the hydrological variable in South Korea in order to improve our capacity to model the hydrological variables. The current study focused on estimating parameters of regional copula and regional marginal models, selecting the most appropriate distribution models, and estimating regional multivariate growth curve in the multivariate regional frequency analysis. Annual maximum rainfall and duration data observed at 71 stations were used for the analysis. The results of the current study indicate that Frank and Gumbel copula models were selected as the most appropriate regional copula models for the employed regions. Several distributions, e.g. Gumbel and log-normal, were the representative regional marginal models. Based on relative root mean square error of the quantile growth curves, the multivariate regional frequency analysis provided more stable and accurate quantiles than the multivariate at-site frequency analysis, especially for long return periods. Application of regional frequency analysis in bivariate rainfall-duration analysis can provide more stable quantile estimation for hydraulic infrastructure design criteria and accurate modelling of rainfall-duration relationship.

• KIISE Transactions on Computing Practices
• /
• v.22 no.8
• /
• pp.375-380
• /
• 2016

The proliferation of online social networking services (OSNSs) and smartphones has enabled people to easily make friends with a large number of users in the online communities, and interact with each other. This leads to an increase in the usage rate of OSNSs. However, individuals who have immersed into their digital lives, prioritizing the virtual world against the real one, become more and more isolated in the physical world. Thus, their socialization processes that are undertaken only through lots of face-to-face interactions and trial-and-errors are apt to be neglected via 'Add Friend' kind of functions in OSNSs. In this paper, we present a friend recommendation system based on the on/off-line contextual information for the OSNS users to have more serendipitous offline interactions. In order to accomplish this, we modeled both offline information (i.e., place visit history) collected from a user's smartphone on a 3D tensor, and online social data (i.e., friend relationships) from Facebook on a matrix. We then recommended like-minded people and encouraged their offline interactions. We evaluated the users' satisfaction based on a real-world dataset collected from 43 users (12 on-campus users and 31 users randomly selected from Facebook friends of on-campus users).

• Korean Journal of Remote Sensing
• /
• v.31 no.6
• /
• pp.513-521
• /
• 2015

In the processing of ocean color remote sensing data, spatio-temporal binning is crucial for securing effective observation area. The validity determination for given source data refers to the information in Level-2 flag. For minimizing the stray light contamination, NASA OBPG's standard algorithm suggests the use of large filtering window but it results in the loss of effective observation area. This study is aimed for quality improvement of ocean color remote sensing data by recovering/extending the portion of effective observation area. We analyzed the difference between MODIS/Aqua standard and modified product in terms of chlorophyll-a concentration, spatial and temporal coverage. The recovery fractions in Level-2 swath product, Level-3 daily composite product, 8-day composite product, and monthly composite product were $13.2({\pm}5.2)%$, $30.8({\pm}16.3)%$, $15.8({\pm}9.2)%$, and $6.0({\pm}5.6)%$, respectively. The mean difference between chlorophyll-a concentrations of two products was only 0.012%, which is smaller than the nominal precision of the geophysical parameter estimation. Increase in areal coverage also results in the increase in temporal density of multi-temporal dataset, and this processing gain was most effective in 8-day composite data. The proposed method can contribute for the quality enhancement of ocean color remote sensing data by improving not only the data productivity but also statistical stability from increased number of samples.

• Journal of Korea Trade
• /
• v.24 no.8
• /
• pp.153-170
• /
• 2020

Purpose - As the US-China trade war intensifies and lasts long time, there is growing concern about its potential effects on the global economy. In particular, for the countries like Korea that have a large economic dependence on the economy of the two countries, the US-China trade war may have a great repercussion in many ways. The aim of this paper is to investigate the global productivity and market structure implications of the US-China trade war for Korea, as well as for other surrounding countries and regions. Design/methodology - In this paper, we develop a full multi-country/region multi-sector computable general equilibrium (CGE) model of global trade incorporating heterogeneous workers and firms in individual skill levels and used technologies. We then calibrate the model using a global Social Accounting Matrix (SAM) dataset extracted from the recently released GTAP 10 Database, and assess the potential effects of the US-China trade war on the aggregate real productivity and the market structure for Korea, as well as for other surrounding countries and regions. Findings - We show that the US-China trade war may largely affect the aggregate productivity in each sector in each country/region, as well as the global market structure through entry and exit of firms, which results finally in considerable changes in the industrial comparative advantage of each country/region. Though the effects are diverse sector by sector, the results show that Korea may also be affected significantly: concerning the real productivity implications, it is shown that the machinery industry may be affected the most negatively; on the other hand, it is shown that the number of exporting firms may decrease the most in the other transports industry. Originality/value - As the US-China trade war intensifies, many studies have tried to estimate the possible implications, and for this usually the CGE models have largely been used as the standard tool for evaluating the impacts of changes in trade policies. Standard CGE models, however, cannot be used to assess the global productivity and market structure implications due to the symmetric and simplified base assumptions. This paper is the first to analyze and quantify the possible impacts of the US-China trade war on the aggregate productivity and global market structure using a CGE model incorporating endogenous skill-technology assignment of heterogeneous workers and firms.

• Journal of the Korea Society of Computer and Information
• /
• v.26 no.11
• /
• pp.41-49
• /
• 2021

All application programs, including malware, call the Application Programming Interface (API) upon execution. Recently, using those characteristics, attempts to detect and classify malware based on API Call information have been actively studied. However, datasets containing API Call information require a large amount of computational cost and processing time. In addition, information that does not significantly affect the classification of malware may affect the classification accuracy of the learning model. Therefore, in this paper, we propose a method of extracting a essential feature set after reducing the dimensionality of API Call information by applying various feature selection methods. We used CICAndMal2020, a recently announced Android malware dataset, for the experiment. After extracting the essential feature set through various feature selection methods, Android malware classification was conducted using CNN (Convolutional Neural Network) and the results were analyzed. The results showed that the selected feature set or weight priority varies according to the feature selection methods. And, in the case of binary classification, malware was classified with 97% accuracy even if the feature set was reduced to 15% of the total size. In the case of multiclass classification, an average accuracy of 83% was achieved while reducing the feature set to 8% of the total size.

• KIPS Transactions on Computer and Communication Systems
• /
• v.10 no.7
• /
• pp.191-198
• /
• 2021

Since the size of training dataset become large and the model is getting deeper to achieve high accuracy in deep learning, the deep neural network training requires a lot of computation and it takes too much time with a single node. Therefore, distributed deep learning is proposed to reduce the training time by distributing computation across multiple nodes. In this study, we propose hybrid allreduce strategy that considers the characteristics of each layer and communication and computational overlapping technique for synchronization of distributed deep learning. Since the convolution layer has fewer parameters than the fully-connected layer as well as it is located at the upper, only short overlapping time is allowed. Thus, butterfly allreduce is used to synchronize the convolution layer. On the other hand, fully-connecter layer is synchronized using ring all-reduce. The empirical experiment results on PyTorch with our proposed scheme shows that the proposed method reduced the training time by up to 33% compared to the baseline PyTorch.