• Proceedings of the KSPS conference
• /
• 2007.05a
• /
• pp.327-329
• /
• 2007

This paper presents a FSN-based LVCSR system and it's application to the speech TV program guide. Unlike the most popular statistical language model-based system, we used FSN grammar based on the graph theory-based FSN optimization algorithm and knowledge-based advanced word boundary modeling. For the memory and latency efficiency, we implemented the dynamic pruning scheduling based on the histogram of active words and their likelihood distribution. We achieved a 10.7% word accuracy improvement with 57.3% speedup.

• The Korean Journal of Applied Statistics
• /
• v.25 no.4
• /
• pp.661-668
• /
• 2012

For displaying multivariate numerical data on a 2D plane by the projection, principal components biplot and the GGobi are two main tools of data visualization. The biplot is very useful for capturing the global shape of the dataset, by representing $n$ observations and $p$ variables simultaneously on a single graph. The GGobi shows a dynamic movie of the images of $n$ observations projected onto a sequence of unit vectors floating on the $p$-dimensional sphere. Even though these two methods are certainly very valuable, there are drawbacks. The biplot is too condensed to describe the detailed parts of the data, and the GGobi is too burdensome for ordinary data analyses. In this paper, "the local projective display(LPD)" is proposed for visualizing multivariate numerical data. Main steps of the LDP are 1) $k$-means clustering of the data into $k$ subsets, 2) drawing $k$ principal components biplots of individual subsets, and 3) sequencing $k$ plots by Hurley's (2004) endlink algorithm for cognitive continuity.

• The Korean Journal of Applied Statistics
• /
• v.25 no.4
• /
• pp.563-573
• /
• 2012

Various methods of analysis have been proposed to understand the gene-disease relation and gene-gene interaction effect for a disease through comparison of genotype in case-control study. In this study, we proposed the method to detect a genetic association and gene-gene interaction through the use of a network graph and centrality measures that are used in social network analysis. The applicability of the proposed method was studied through an analysis of real genetic data.

• Proceedings of the IEEK Conference
• /
• summer
• /
• pp.259-262
• /
• 2004

In most of the character recognition systems, the method of template matching or statistical method using hidden Markov model is used to extract and recognize feature shapes. In this paper, we used modified chain-code which has 8-directions but 4-codes, and made the chain-code of hand-written character, after that, converted it into transition chain-code by applying to HMM(Hidden Markov Model). The transition chain code by HMM is analyzed as signal flow graph by Mason's theory which is generally used to calculate forward gain at automatic control system. If the specific forward gain and feedback gain is properly set, the forward gain of transition chain-code using Mason's theory can be distinguished depending on each object for recognition. This data of the gain is reorganized as tree structure, hence making it possible to distinguish different hand-written characters. With this method, $91\%$ recognition rate was acquired.

• Journal of the Korean Data and Information Science Society
• /
• v.9 no.1
• /
• pp.11-18
• /
• 1998

In clinical measurement comparison of a new measurement technique with an established one is often needed to see whether they agree sufficiently for the new to replace the old. Such investigations are often analysed inappropriately, notably by using the correlation coefficient(r). So, the measurement for agreement was determined by Bland & Altman's method recently. In this article, we will analyse the measurement for agreement by using Q-Q plot and by applying Bland and Altman's method through graph. And we will show characteristics for these techniques.

• Journal of the Institute of Electronics Engineers of Korea SD
• /
• v.43 no.8 s.350
• /
• pp.35-43
• /
• 2006

In the paper, we propose a new method for power optimization that uses path balancing based on stochastic estimation of glitch in Statistical Static Timing Analysis (SSTA). The proposed method estimates the probability of glitch occurrence using tightness probability of each node in timing graph. In addition, we propose efficient gate sizing technique for glitch reduction using accurate calculation of sizing effect in delay considering probability of glitch occurrence. The efficiency of proposed method has been verified on ISCAS85 benchmark circuits with $0.16{\mu}m$ model parameters. Experimental results show up to 8.6% of accuracy improvement in glitch estimation and 9.5% of optimization improvement.

• Journal of Information Technology Applications and Management
• /
• v.22 no.3
• /
• pp.65-81
• /
• 2015

This study reports a preliminary finding of the types and numbers of graphs being presented in the annual reports of about thirty top listed companies trading publicly in the stock markets of three countries-Thailand (SET), Malaysia (BM), and Singapore (SGX)-that were chosen based on their inclusion in the ASEAN Stars Index under the ASEAN Trading Link project. A total of 6,753 graphs from nineteen sectors were extracted and examined. Banking, real estate, and telecommunications are ranked the three most condense sectors, accounting for 50.2% of the total number of graphs observed. The three most used graphs are the Conservative Bar, Donut graph and Stack Bar. Less than one percent of Infographic type graphs were used. The five most depicted graphed variables are Asset, Revenue, Net profit, Liability, and Dividend. Using rudimentary framework to detect distorted or misleading statistical graphs, the study found 60.6% of the graphs distorted across the three markets, SET, BM, and SGX. BM ranked first in percentages of graphs being distortedly presented (73%). The other two markets, SET and SGX, have about the same proportions, 53.88% and 53.03%, respectively. Likewise, the proportions of Well-designed versus Inappropriate-designed graphs of the latter two markets are a little over one time (SET = 1 : 1.17; SGX = 1 : 1.13), whereas the proportion is almost triple for the BM market (BM = 1 : 2.70). In addition, the trend of distorted graphs found is slightly increasing as the longevity of the ASEAN Stars Index increases. One possible explanation for the relatively equal proportion of inappropriate graphs found is that SET is the smallest market and SGX, though the largest, is the most regulated market. BM, on the other hand, may want to present their financial data in the most attractive manner to prospective investors, thus, regulatory constraints and governance structure are still lenient.

• Journal of KIISE:Information Networking
• /
• v.36 no.2
• /
• pp.69-79
• /
• 2009

Today's Internet is one of the necessaries of our life. Anomalies of the Internet provoke social problems. For that reason, Internet Measurement which studies characteristics on Internet traffic attracts pubic attention. Recently, Traffic Dispersion Graph (TDG), a novel traffic analysis method, was proposed. The TDG is not a statistical analysis method but a graphical visualization method on interactions among network components. In this paper, we propose a new anomaly detection paradigm and its technique using TDG. The existing studies have focused on detecting anomalous packets of flows. On the other hand, we focus on detecting the sources of anomalous traffic. To realize our paradigm, we designed the TDG Clustering method. Through this method, we could classify anomalous hosts infected by various worm viruses. We obtained normal traffic through dropping traffic of the anomalous hosts. Especially, we expect that the TDG clustering method can be applied to real-time anomaly detection because calculations of the method are fast.

• The Korean Journal of Applied Statistics
• /
• v.30 no.3
• /
• pp.403-415
• /
• 2017

In this paper, we investigate cosponsorship networks found in the 17th National Assembly of Republic of Korea. New legislation should be sponsored by at least 10 legislators including one main sponsor. Cosponsorship networks can be constructed, using directional links from cosponsors of legislation to its main sponsor; subsequently, these networks indicate the social relationships among the legislators. We apply Exponential Random Graph Model (ERGM) for valued networks to capture structural properties and the covariate effects of networks. We find the effect of the same party has the greatest influence on the composition of the network. Mutuality also plays an important role in the cosponsorship network; in addition, the effect of the number of elections won by a legislator has a small but significant influence.

• The Korean Journal of Applied Statistics
• /
• v.30 no.6
• /
• pp.943-955
• /
• 2017

We investigate and estimate the causal effect of the survey methods in telephone surveys for the 19th presidential election. For this causal study, we draw a causal graph that represents the causal relationship between variables. Then we decide which variables should be included in the model and which variables should not be. We explain why the research agency is a should-be variable and the response rate is a shouldnot-be variable. The effect of ARS can not be estimated due to data limitations. We have found that there is no significant difference in the effect of the proportion of cell phone survey if it is less than about 90 percent. But the support rate for Moon Jae-in gets higher if the survey is performed only by cell phones.