• The Journal of Korea Institute of Information, Electronics, and Communication Technology
• /
• v.17 no.2
• /
• pp.113-120
• /
• 2024

The population sparseness index(PSI) is being utilized to describe the functioning of internal layers in artificial neural networks from the perspective of neurons, shedding light on the black-box nature of the network's internal operations. There is research indicating a positive correlation between the PSI and performance in each layer of convolutional neural network models for image classification. In this study, we observed the internal operations of a convolutional neural network when adversarial examples were applied. The results of the experiments revealed a similar pattern of positive correlation for adversarial examples, which were modified to maintain 5% accuracy compared to applying benign data. Thus, while there may be differences in each adversarial attack, the observed PSI for adversarial examples demonstrated consistent positive correlations with benign data across layers.

• Journal of the Korean earth science society
• /
• v.31 no.4
• /
• pp.301-312
• /
• 2010

Geochemical data have been regarded as one of the important environmental variables in the environmental management. Since they are often sampled at sparse locations, it is important not only to predict attribute values at unsampled locations, but also to assess the uncertainty attached to the prediction for further analysis. The main objective of this paper is to exemplify how indicator geostatistics can be effectively applied to geochemical data processing for providing decision-supporting information as well as spatial distribution of the geochemical data. A whole geostatistical analysis framework, which includes probabilistic uncertainty modeling, classification and risk analysis, was illustrated through a case study of cadmium mapping. A conditional cumulative distribution function (ccdf) was first modeled by indicator kriging, and then e-type estimates and conditional variance were computed for spatial distribution of cadmium and quantitative uncertainty measures, respectively. Two different classification criteria such as a probability thresholding and an attribute thresholding were applied to delineate contaminated and safe areas. Finally, additional sampling locations were extracted from the coefficient of variation that accounts for both the conditional variance and the difference between attribute values and thresholding values. It is suggested that the indicator geostatistical framework illustrated in this study be a useful tool for analyzing any environmental variables including geochemical data for decision-making in the presence of uncertainty.

• Journal of the Korean earth science society
• /
• v.36 no.1
• /
• pp.36-50
• /
• 2015

The purpose of this study is to assign emission source profiles of volatile organic compounds (VOCs) and particulate matters (PMs) for chemical speciation, and to correct the temporal allocation factor and the chemical speciation of source profiles according to the source classification code within the sparse matrix operator kernel emission system (SMOKE) in the Seoul metropolitan area. The chemical speciation from the source profiles of VOCs such as gasoline, diesel vapor, coating, dry cleaning and LPG include 12 and 34 species for the carbon bond IV (CBIV) chemical mechanism and the statewide air pollution research center 99 (SAPRC99) chemical mechanism, respectively. Also, the chemical speciation of PM2.5 such as soil, road dust, gasoline and diesel vehicles, industrial source, municipal incinerator, coal fired, power plant, biomass burning and marine was allocated to 5 species of fine PM, organic carbon, elementary carbon, $NO_3{^-}$, and $SO_4{^2-}$. In addition, temporal profiles for point and line sources were obtained by using the stack telemetry system (TMS) and hourly traffic flows in the Seoul metropolitan area for 2007. In particular, the temporal allocation factor for the ozone modeling at point sources was estimated based on $NO_X$ emission inventories of the stack TMS data.

• The Journal of Korean Academy of Prosthodontics
• /
• v.28 no.2
• /
• pp.91-108
• /
• 1990

The purpose of this study was to compare the maganitude of the discrepancies of the mounting errors in according to the states of dentitions, and to the superoinferior, anteroposterior, and rightleft driecetions. GROUP I. : Fourteen patients 22 to 26 years of age with a full complement of teeth, were used in the study. The criteria fro patient selection were a complete dentition, sparse restorarive treatment, and adequate posterior and anterior occlusan stops. And they had no sign and sympton at TMG area. GROUPII. : Eigth patients 37 to 62 years of age with bilateral free ends. The criteria for patient selection were Kennedy classification class 1 cases, and adequate posterior and anterior stops. And the opposite dentitions were a full complement of teeth. Irreversible hydrocolloid impresiion of each arch was taken of each patient. These were immediatel poured in stone and mounted on a Denar Mark II. Articulator with the arbitrary slidematic face-bow. With hand articulation th e mandibular cast was mounted to the maxillary cast in centric occlusion. Five types of interocclusal records were taken of each patient : (1) aluwax (2) baseplate wax; (3) znic oxide-eugenol pasts; (4) polyether (Ramitec); (5) modeling compound. All measurement of the five selected recording materials were compared with those of the hand-articulated full arch models in centric occlusion or maximum interdigitation. The results were as follows; 1. There were statistical differences in amount of devitation in according to the materials and the states of dentition. The amount of deviation of compound was the largest. 2. There were statistical differences in amount of deviation in complete dentition at all directions. The amount of diviation of compound was the largest. And at the right-left direction the amount of znic oxide-eugenol paste was larger than that of baseplate wax. 3. There was a statistical difference in amount of diviations in partial edentulous dentition at the superoinferior direction. The amount of deviation of compound was larger than that of znic oxide-eugenol paste. 4. There was as statistical difference in amount of deviations in partial edentulous dentition at the right-left direction. The amount of deviation of baseplate wax was larger that tnat of polyether. 5. There was not a statistical difference in amount of diviation in partial edentulous dentition at the anteroposterior direction.

• Korean Journal of Plant Taxonomy
• /
• v.38 no.2
• /
• pp.163-177
• /
• 2008

Pollen grains of 17 taxa (14 species with two additional subspecies and one variety) of genera Patrinia Juss. and Nardostachys DC. in tribe Patrinieae (Valerianaceae) were studied using light microscope, scanning electron microscope, and transmission electron microscope. Pollen grains are medium ($37.41{\times}43.60{\mu}m$ - $45.65{\times}48.50{\mu}m$) to large ($54.88{\times}59.41{\mu}m$ - $61.70{\times}71.00{\mu}m$) in size, tricolpate (rarely tetracolpate) with the characteristic halo surrounding the aperture. In equatorial view, the pollen is oblate to subprolate, and in polar view, it is mostly circular or rarely 3-lobed. Two major pollen types are recognized on the basis of exine sculpturing patterns; Type I: Exine is composed of echinae together with sparse or dense microechinae, and verrucae shallow or rarely absent (Nardostachys and sections Paleopatrinia, and Monanadropatrinia of Patirinia). - Type II: Exine is composed of massive echinae together with dense microechinae, and prominent verrucae (section Centrotrinia of Patrinia). In TEM sections, columellae are extended from the footlayer into verrucae, and exine thickness is uniform at the pole and equator. Additionally, the infrageneric classification systems of the Patrinieae were evaluated on the basis of the present data.

• Journal of the Korea Institute of Information Security & Cryptology
• /
• v.21 no.6
• /
• pp.183-190
• /
• 2011

In this paper, we propose a prototype system for abnormal sound detection and identification which detects and recognizes the abnormal situations by means of analyzing audio information coming in real time from CCTV cameras under surveillance environment. The proposed system is composed of two layers: The first layer is an one-class support vector machine, i.e., support vector data description (SVDD) that performs rapid detection of abnormal situations and alerts to the manager. The second layer classifies the detected abnormal sound into predefined class such as 'gun', 'scream', 'siren', 'crash', 'bomb' via a sparse representation classifier (SRC) to cope with emergency situations. The proposed system is designed in a hierarchical manner via a mixture of SVDD and SRC, which has desired characteristics as follows: 1) By fast detecting abnormal sound using SVDD trained with only normal sound, it does not perform the unnecessary classification for normal sound. 2) It ensures a reliable system performance via a SRC that has been successfully applied in the field of face recognition. 3) With the intrinsic incremental learning capability of SRC, it can actively adapt itself to the change of a sound database. The experimental results with the qualitative analysis illustrate the efficiency of the proposed method.

• Journal of Intelligence and Information Systems
• /
• v.23 no.2
• /
• pp.123-138
• /
• 2017

Since the stock market is driven by the expectation of traders, studies have been conducted to predict stock price movements through analysis of various sources of text data. In order to predict stock price movements, research has been conducted not only on the relationship between text data and fluctuations in stock prices, but also on the trading stocks based on news articles and social media responses. Studies that predict the movements of stock prices have also applied classification algorithms with constructing term-document matrix in the same way as other text mining approaches. Because the document contains a lot of words, it is better to select words that contribute more for building a term-document matrix. Based on the frequency of words, words that show too little frequency or importance are removed. It also selects words according to their contribution by measuring the degree to which a word contributes to correctly classifying a document. The basic idea of constructing a term-document matrix was to collect all the documents to be analyzed and to select and use the words that have an influence on the classification. In this study, we analyze the documents for each individual item and select the words that are irrelevant for all categories as neutral words. We extract the words around the selected neutral word and use it to generate the term-document matrix. The neutral word itself starts with the idea that the stock movement is less related to the existence of the neutral words, and that the surrounding words of the neutral word are more likely to affect the stock price movements. And apply it to the algorithm that classifies the stock price fluctuations with the generated term-document matrix. In this study, we firstly removed stop words and selected neutral words for each stock. And we used a method to exclude words that are included in news articles for other stocks among the selected words. Through the online news portal, we collected four months of news articles on the top 10 market cap stocks. We split the news articles into 3 month news data as training data and apply the remaining one month news articles to the model to predict the stock price movements of the next day. We used SVM, Boosting and Random Forest for building models and predicting the movements of stock prices. The stock market opened for four months (2016/02/01 ~ 2016/05/31) for a total of 80 days, using the initial 60 days as a training set and the remaining 20 days as a test set. The proposed word - based algorithm in this study showed better classification performance than the word selection method based on sparsity. This study predicted stock price volatility by collecting and analyzing news articles of the top 10 stocks in market cap. We used the term - document matrix based classification model to estimate the stock price fluctuations and compared the performance of the existing sparse - based word extraction method and the suggested method of removing words from the term - document matrix. The suggested method differs from the word extraction method in that it uses not only the news articles for the corresponding stock but also other news items to determine the words to extract. In other words, it removed not only the words that appeared in all the increase and decrease but also the words that appeared common in the news for other stocks. When the prediction accuracy was compared, the suggested method showed higher accuracy. The limitation of this study is that the stock price prediction was set up to classify the rise and fall, and the experiment was conducted only for the top ten stocks. The 10 stocks used in the experiment do not represent the entire stock market. In addition, it is difficult to show the investment performance because stock price fluctuation and profit rate may be different. Therefore, it is necessary to study the research using more stocks and the yield prediction through trading simulation.