• 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.

• Microbiology and Biotechnology Letters
• /
• v.8 no.3
• /
• pp.181-191
• /
• 1980

For the purpose of studies on the citric acid production, some experiments were carried out with isolated strains. The results obtained were as follows. 1) The optimal culture media of the strain M-80 in surface culture contained 140g of sucrose, 3.0g of (N $H_4$)$_2$S $O_4$, 1.5g of K $H_2$P $O_4$, 0.2g of MgS $O_4$.7$H_2O$, 3.0mg of F $e^{++}$, 1.0mg of Z $n^{++}$, 0.5N HCI to a pH of 5.0 and distilled water to 1.0 liter; and that of the strain M-315 in surface culture contained 140g of sucrose, 2.0g of N $H_4$N $O_3$, 1.0g of K $H_2$P $O_4$, 0.25g of MgS $O_4$. 7$H_2O$, 2.0mg of F $e^{++}$, 2.0mg of Z $n^{++}$, 0.05mg of C $u^{++}$, 0.5N HCI to a pH of 4.5 and distilled water to 1.0 liter. While that of the strain M-315 in submerged culture contained 140g of sucrose, 2.5g of N $H_4$N $O_3$, 1.5g of K $H_2$P $O_4$, 0.3g of MgS $O_4$. 7$H_2O$, 3.0mg of F $e^{++}$, 0.1mg of C $u^{++}$, 0.5N HCI to a pH of 4.5 and distilled water to 1.0 liter. The optimal temperature and size of inoculum were mostly 28-3$0^{\circ}C$, 10$^{7}$ -10$^{8}$ spores/50ml, respectively. 2) Through the course of citric acid production, the growth of strains had nearly been completed, pH value was rapidly decreased below 2.0 and the content of sugar was also reduced, while the accumulation of citric acid in media was remarkably begun in about 3-4 days. The yields of citric acid generally reached the maximum level in 8-10 days in surface or submerged fermentation process. 3) Methanol was effective citric acid production when they were added to fermentation media. In the case of surface culture, by addition of 2％ (strain M-80), 3％ (strain M-315), the yields of citric acid was increased 6.5％, 20.6%, respectively and 5.0% yield was increased by addition of 3% methanol in submerged culture media of the strain M-315. 4) Chromatography analysis of culture broth after fermentation under optimal culture conditions detected that the majority of acid in media was citric acid. 72.1mg/ml, 98.1mg/ml, of citric acid were determined in surface culture media by strains of M-80, M-315, and 59.8 mg/ml of citric acid was contained in the submerged culture media by the strain M-315. strain M-315.

• Journal of Animal Science and Technology
• /
• v.47 no.6
• /
• pp.975-984
• /
• 2005

This study fulfilled to investigate the feed efficiency, tissue selenium distribution, carcass characteristic and economic value in finishing Hanwoo steers fed organic selenium mix (OSM) which included seleno-yeast, rumen culture and other microbial supplements. Forty five finishing Hanwoo steers were tested for 4 months dividing to three feeding groups: OSM add as 0.5 ppm Se of DM feeds (0.5 ppm OSM), OSM enriched add as 1.0 ppm Se of DM feeds (1.0 ppm OSM) and basal diet without OSM (control). The total weight gains, the average daily gains and the feed intakes were not differ in treatments (p > 0.05). No differences (p > 0.05) were noted for hot carcass weight, loin eye area, backfat thickness, meat yield index, meat color, fat color, tenderness and maturity. However, the 1.0 ppm OSM showed better performances for feed requirement, TDN per gain, meat yield grade and meat quality grade compared to other groups. Tissue selenium distribution was increased by organic selenium feeding: higher Se concentration in liver and rump of 0.5 ppm OSM (p < 0.05), and kidney, liver, sirloin and rump of 1.0 ppm OSM (p < 0.05) than the tissues of control group. Generally, tissue selenium was the highest value in 1.0 ppm OSM and showed higher concentrate in order; kidney, liver, sirloin and rump. The income over feed cost was 1.06-fold higher in 1.0 ppm OSM than control group. In conclusion, organic selenium mix supplementation and its amounts were not influenced to feed intake, body gain and carcass characteristic but significantly increased tissue selenium. Therefore, these results suggest that finishing Hanwoo steer fed an enriched organic selenium mix with proper probiotics is able to produce “high-Se” beef as high bioavailable form as well as create a beneficial opportunity on Hanwoo farm.

• Journal of Animal Environmental Science
• /
• v.14 no.2
• /
• pp.119-128
• /
• 2008

This experiment was conducted to investigate the effects of concentrated pig slurry using membrane filter on growth of tomato in nutriculture. Pig slurry was filtered by ultra filtration and concentrated by reverse osmosis process. Filtration of pig slurry was necessary to prevent the hose clogging in nutriculture. The concentrated pig slurry (CS) and nutrient solution (NS) were mixed by six different mixing ratios of 0:100, 20:80, 40:60, 60:40%, 80: 20 and 100%:0% based on nitrogen content. The chemical nutrient solution was the solution of National Horticulture Research Station for the growth of tomato. The concentration of nutrient solution was adjusted a range of $1.6{\sim}2.0mS/cm$ in EC. The plant height of tomato treated with CS 20+NS 80% was similar with NS 100% control plot. Plant height was highest in the plot of CS 20+NS 80%. The treatment of 100% concentrated pig slurry was lowest in the gowth characteristics of tomato. Number of cluster was very lower in 100% concentrated pig slurry compared with plot of chemical nutrient solution. In the beginning of growth stage, SPAD reading value was reduced in plot treated with CS 100%, but CS 20+NS 80% plot was higher compared to 100% concentrated pig slurry. SPAD value of tomato leaves was decreased as the amount of CS was increased. The SPAD value also in treatment of concentrated pig slurry was lower in the middle growth stage compared to control plot. The dry weight of stem and leaf were 107.4, 104.2g in plot of NS 100% and CS 20%+NS 80%, respectively. The fruit number and weight were decreased at high application plots of concentrated pig slurry, The fruit setting of tomato showed lowest in the plot treated with 100% concentrated pig slurry, and the growth of tomato severely decreased after application of 100% CS treatment. In conclusion, the growth characteristics such as plant height and fruit weight of tomato were not significantly different between the plots treated with mixture of 20% CS +80%NS and 100% nutrient solution treatment. In conclusion, the mixture solution of 20% of concentrated pig slurry and 80% of nutrient solution could be used as a nutrition solution of tomato nuticulture.

• Korean journal of food and cookery science
• /
• v.11 no.4
• /
• pp.356-361
• /
• 1995

This study was conducted to know the quality of marketed green teas which were devided into high, middle and low grade according to the price. we examined the content of water, ash, total nitrogen, tannin, caffeine, chlorophyll and color value and did sensory evaluation. The results were as follows: 1) The content of water was 2.17-3.67％ and content of ash was 4.50-5.17％, there were no significant difference in each grade. 2) The content of total nitrogen was 4.38％ in high grade, 4.60％ in middle grade and 4.68％ in low grade. 3) The content of tannin was 11.09％ in high grade, 14.22％ in middle grade and 14.44％ in low grade, middle and low grade were significantly higher than high grade. 4) The nitrogen rate to tannin (N/T) was 39.5％ in high grade, 32.6％ in middle grade and 32.70％ in low grade, N/T rate of high grade was significantly highest than others. 5) The content of caffeine is 1.78％ in high grade, 1.32％ in middle grade and 0.92％ in low grade, high and middle grade were significantly higher than low grade. 6) The content of total chlorophyll is 297.37 mg％ in high grade, 192.89 mg％ in middle grade and 204.79 mg％ in low grade, chlorophyll a was 69.23 mg％ in high grade, 51.99 mg％ in middle grade and 63.42 mg％ in low grade, and chlorophyll b was 228.47 mg％ in high grade, 131.21 mg％ in middle grade and 141.63 mg％ in low grade. 7) Yellow value of high and middle grade were significantly higher than low grade, but blue and red value were no significant diffrence in groups. 8) In sensory evaluation, appearance, taste, odor of high grade were better than others and total score was 15.25 in high grade, 12.97 in middle grade and 9.80 in low grade. 9) Price had a positive correlation with caffeine, appearance and taste, but a negative correlation with tannine. NT rate had a negative correlation with tannine, but a positive correlation with caffeine.

• Journal of Intelligence and Information Systems
• /
• v.17 no.1
• /
• pp.111-125
• /
• 2011

Clustering is a process of grouping similar or relevant documents into a cluster and assigning a meaningful concept to the cluster. By this process, clustering facilitates fast and correct search for the relevant documents by narrowing down the range of searching only to the collection of documents belonging to related clusters. For effective clustering, techniques are required for identifying similar documents and grouping them into a cluster, and discovering a concept that is most relevant to the cluster. One of the problems often appearing in this context is the detection of a complex concept that overlaps with several simple concepts at the same hierarchical level. Previous clustering methods were unable to identify and represent a complex concept that belongs to several different clusters at the same level in the concept hierarchy, and also could not validate the semantic hierarchical relationship between a complex concept and each of simple concepts. In order to solve these problems, this paper proposes a new clustering method that identifies and represents complex concepts efficiently. We developed the Hierarchical Overlapping Clustering (HOC) algorithm that modified the traditional Agglomerative Hierarchical Clustering algorithm to allow overlapped clusters at the same level in the concept hierarchy. The HOC algorithm represents the clustering result not by a tree but by a lattice to detect complex concepts. We developed a system that employs the HOC algorithm to carry out the goal of complex concept detection. This system operates in three phases; 1) the preprocessing of documents, 2) the clustering using the HOC algorithm, and 3) the validation of semantic hierarchical relationships among the concepts in the lattice obtained as a result of clustering. The preprocessing phase represents the documents as x-y coordinate values in a 2-dimensional space by considering the weights of terms appearing in the documents. First, it goes through some refinement process by applying stopwords removal and stemming to extract index terms. Then, each index term is assigned a TF-IDF weight value and the x-y coordinate value for each document is determined by combining the TF-IDF values of the terms in it. The clustering phase uses the HOC algorithm in which the similarity between the documents is calculated by applying the Euclidean distance method. Initially, a cluster is generated for each document by grouping those documents that are closest to it. Then, the distance between any two clusters is measured, grouping the closest clusters as a new cluster. This process is repeated until the root cluster is generated. In the validation phase, the feature selection method is applied to validate the appropriateness of the cluster concepts built by the HOC algorithm to see if they have meaningful hierarchical relationships. Feature selection is a method of extracting key features from a document by identifying and assigning weight values to important and representative terms in the document. In order to correctly select key features, a method is needed to determine how each term contributes to the class of the document. Among several methods achieving this goal, this paper adopted the $x^2$�� statistics, which measures the dependency degree of a term t to a class c, and represents the relationship between t and c by a numerical value. To demonstrate the effectiveness of the HOC algorithm, a series of performance evaluation is carried out by using a well-known Reuter-21578 news collection. The result of performance evaluation showed that the HOC algorithm greatly contributes to detecting and producing complex concepts by generating the concept hierarchy in a lattice structure.

• KOMUNHWA
• /
• no.68
• /
• pp.41-57
• /
• 2006

The purpose of this study is to review educational value of cultural heritage sites first and then, to examine the importance of interpretation and interpretation methods and process briefly to utilize cultural heritage sites as important resources for public education and cultural tourism. The educational value of cultural heritage sites is not limited to provide visitors with various information on historical facts and events related to the cultural heritage site, but to provide the public with a new recognition of the importance of the preservation of cultural heritage sites by understanding the site well. In order to utilize the educational value of the cultural heritage site, interpretation must be emphasized as an important element of cultural heritage sites management and much attention should be given on the development and practice of interpretation programs in each cultural heritage site. In order to develop an interpretation program of the site, a specific target audience must be defined first to find most resonable scope and direction of an interpretation program and then, site specific interpretation themes (or a theme) should be developed from the interpretive topic. According to these defined elements, an interpretation program(s) is established by using various interpretation methods and techniques in order to provide visitors with useful information and knowledge on the site efficiently. Interpretation methods and techniques are various as defined by various ways and most effective methods or techniques are employed as a specific interpretation program is presented most effectively. The evaluation processes in planing stage, during developing stage, and after completion of interpretation program development are important processes to find out any problems of planed interpretation program and to revise or update the interpretation program. Conclusively, the direction of cultural heritage sites management in Korea has focused on the conservation and restoration of cultural heritage sites to preserve the original conditions of cultural heritage sites so far. This direction on current practice on cultural heritage sites management should he changed to more active utilization of the sites and interpretation of the sites is a way to accomplish the this goal.

• Journal of Korean Society of Forest Science
• /
• v.90 no.3
• /
• pp.295-305
• /
• 2001

This study was conducted to reveal stand structure and competition status by site types of natural deciduous forest in Pyungchang, Kangwon-do. The study site was divided by nine different types based on elevation(higher than 1,000m, 700~1,000m. lower than 700m) and topography(ridge, slope, valley). The objective of this study is to provide a basic information necessary for the environment-friendly management methods of natural deciduous forest on the basis of the stand structure and competition status by site types. It is confirmed that the range from 700m to 1,000m in elevation was the most suitable for stand growth. Species diversity and richness index also tended to be increased as elevation decreased. The ridge was the best in terms of stand growth by topography and followed by slope and valley in order. In addition, as expected, species with high importance value showed relatively low evenness index. The distance-independent competition index was selected as the best competition index model in seven site types of natural deciduous forest. On the other hand, the distance-dependent competition index was highly correlated with periodic annual increment of diameter in both ridge at higher than 1,000m and valley of 700m to 1,000m in elevation. It is proved, as a result, that the best competition index model is somewhat different by site types. From the analysis growth characteristics and competition status by site types, it is identified that the species with high importance value performed well in both growth and competition. The growth of Q. mongolica was excellent in the areas of higher elevation than 700m. Although K. pilus had relatively low importance value in higher elevation than 1,000m and ridge of 700m to 1,000m, the species had stronger competition status rather than other species. Also, U. davidiana and M. amurensis were good in competition status at lower elevation than 1,000m. It is necessary, therefore, that appropriate tending practice should be adopted based on the growth pattern and competition status of each species distributed by site types.

• The Journal of Korean Society for Radiation Therapy
• /
• v.29 no.1
• /
• pp.27-35
• /
• 2017

Purpose: On the left side, breast cancer patients have more side effects than those on the right side because of unnecessary doses in normal organs such as heart and lung. DIBH is performed to reduce this. To evaluate the dose of peripheral organs in the left breast cancer including supraclavicular lymph nodes and internal mammary lymph nodes according to the treatment planning method of Conventional Radiation Therapy, Intensity Modulated Radiation Therapy and Volumetric Modulated Arc Therapy. Materials and Methods: We performed CT-simulation using free breathing and deep inspiration breath-hold technique for 8 patients including left supraclavicular lymph nodes and internal mammary lymph nodes. Based on the acquired CT images, the contour of the body is drawn and the convention is performed so that $95%{\leftarrow}PTV$, $Dmax{\leftarrow}110%$. Conventional Radiation Therapy used a one portal technique on the supraclavicular lymph node and used a field in field technique tangential beam on the breast. Intensity Modulated Radiation Therapy was composed of 7 static fields. Volumetric Modulated Arc Therapy was planned using 2 ARC with a turning radius of $290^{\circ}$ to $179^{\circ}$. The peripheral normal organs dose was analyzed by referring to the dose volume of Eclipse. Results: By applying the deep inspiration breath-hold technique, the mean interval between the heart and chest wall increased $1.6{\pm}0.6cm$. The mean dose of lung was $19.2{\pm}1.0Gy$, which was the smallest value in Intensity Modulated Radiation Therapy. The V30 (%) of the heart was $2.0{\pm}1.9$, which was the smallest value in Intensity Modulated Radiation Therapy. In the left anterior descending coronary artery, the dose was $25.4{\pm}5.4Gy$, which was the smallest in Intensity Modulated Radiation Therapy. The maximum dose value of the Right breast was $29.7{\pm}4.3Gy$ at Intensity Modulated Radiation Therapy. Conclusion: When comparing the values of surrounding normal organs, Intensity Modulated Radiation Therapy and Volumetric Modulated Arc Therapy were applicable values for treatment. Among them, Intensity Modulated Radiation Therapy is considered to be a suitable treatment planning method.

• Restorative Dentistry and Endodontics
• /
• v.25 no.1
• /
• pp.46-55
• /
• 2000

During bonding procedure of composite resin, the prepared cavity can be contaminated by saliva. In this study, marginal microleakage and shear bond strength of a composite resin to primed enamel and dentin treated with artificial saliva(Taliva$^{(R)}$) were evaluated. For the marginal microleakage test, Class V cavities were prepared in the buccal surfaces of fifty molars. The samples were randomly assigned into 5 groups with 10 samples in each group. Control group was applied with a bonding system (Scotchbond$^{TM}$ Multi-Purpose plus) according to manufacture's directions without saliva contamination. Experimental groups were divided into 4 groups and contaminated with artificial saliva for 30 seconds after priming: Experimental 1 group ; artificial saliva was dried with compressed air only, Experimental 2 group ; artificial saliva was rinsed and dried. Experimental 3 group ; cavities were etched with 35% phosphoric acid for 15 seconds after rinsing and drying artificial saliva. Experimental 4 group ; cavities were etched with 35% phosphoric acid for 15 seconds and primer was reapplied after rinsing and drying artificial saliva. All the cavities were applied a bonding agent and filled with a composite resin (Z-100$^{TM}$). Specimens were immersed in 0.5% basic fuschin dye for 24 hours and embedded in transparent acrylic resin and sectioned buccolingually with diamond wheel saw. Four sections were obtained from one specimen. Degree of marginal leakage was scored under stereomicroscope and their scores were averaged from four sections. The data were analyzed by Kruscal-Wallis test and Fisher's LSD. For the shear bond strength test, the buccal or occlusal surfaces of one hundred molar teeth were ground to expose enamel(n=50) or dentin(n=50) using diamond wheel saw and its surface was smoothed with Lapping and Polishing Machine(South Bay Technology Co., U.S.A.). Samples were divided into 5 groups. Treatment of saliva-contaminated enamel and dentin surfaces was same as the marginal microleakage test and composite resin was bonded via a gelatin capsule. All specimens were stored in distilled water for 48 hours. The shear bond strengths were measured by universal testing machine (AGS-1000 4D, Shimaduzu Co., Japan) with a crosshead speed of 5 mm/minute. Failure mode of fracture sites was examined under stereomicroscope. The data were analyzed by ANOVA and Tukey's studentized range test. The results of this study were as follows : 1. Enamel marginal microleakage showed no significant difference among groups. 2. Dentinal marginal microleakages of control, experimental 2 and 4 groups were lower than those of experimental 1 and 3 groups (p<0.05). 3. The shear bond strength to enamel was the highest value in control group (20.03${\pm}$4.47MPa) and the lowest value in experimental 1 group (13.28${\pm}$6.52MPa). There were significant differences between experimental 1 group and other groups (p<0.05). 4. The shear bond strength to dentin was higher in control group (17.87${\pm}$4.02MPa) and experimental 4 group (16.38${\pm}$3.23MPa) than in other groups, its value was low in experimental 1 group (3.95${\pm}$2.51 MPa) and experimental 2 group (6.72${\pm}$2.26MPa)(p<0.05). 5. Failure mode of fractured site on the enamel showed mostly adhesive failures in experimental 1 and 3 groups. 6. Failure mode of fractured site on the dentin did not show adhesive failures in control group, but showed mostly adhesive failure in experimental groups. As a summary of above results, if the primed tooth surface was contaminated with artificial saliva, primer should be reapplied after re-etching it.