• Journal of the Korean Data and Information Science Society
• /
• v.15 no.2
• /
• pp.331-337
• /
• 2004

In the analysis of massive volume data, a neural network model is a useful tool. To implement the Neural network model, it is important to select initial value. Since the initial values are generally used as random value in the neural network, the convergent performance and the prediction rate of model are not stable. To overcome the drawback a possible method use samples randomly selected from the whole data set. That is, coefficients estimated by logistic regression based on the samples are the initial values.

• Communications for Statistical Applications and Methods
• /
• v.7 no.1
• /
• pp.65-86
• /
• 2000

The estimates from an EM when it is applied to a large causal model of 10 or more categorical variables are often subject to the initial values for the estimates. This phenomenon becomes more serious as the model structure becomes more serious as the model structure becomes more complicated involving more variables. In this regard Wu(1983) recommends among others that EMs are implemented several times with different sets of initial values to obtain more appropriate estimates. in this paper a new approach for initial values is proposed. The main idea is that we use initials that are calibrated to data. A simulation result strongly indicates that the calibrated initials give rise to the estimates that are far closer to the true values than the initials that are not calibrated.

• Nuclear Engineering and Technology
• /
• v.55 no.12
• /
• pp.4607-4616
• /
• 2023

We implement machine learning regression models to predict peak pressures of primary and secondary systems, a major safety concern in Loss Of Condenser Vacuum (LOCV) accident. We selected the Multi-dimensional Analysis of Reactor Safety-KINS standard (MARS-KS) code to analyze the LOCV accident, and the reference plant is the Korean Optimized Power Reactor 1000MWe (OPR1000). eXtreme Gradient Boosting (XGBoost) is selected as a machine learning tool. The MARS-KS code is used to generate LOCV accident data and the data is applied to train the machine learning model. Hyperparameter optimization is performed using a simulated annealing. The randomly generated combination of initial conditions within the operating range is put into the input of the XGBoost model to predict the peak pressure. These initial conditions that cause peak pressure with MARS-KS generate the results. After such a process, the error between the predicted value and the code output is calculated. Uncertainty about the machine learning model is also calculated to verify the model accuracy. The machine learning model presented in this paper successfully identifies a combination of initial conditions that produce a more conservative peak pressure than the values calculated with existing methodologies.

• International Journal of Computer Science & Network Security
• /
• v.22 no.1
• /
• pp.167-174
• /
• 2022

In metaheuristic algorithms such as Genetic Algorithm (GA), initial population has a significant impact as it affects the time such algorithm takes to obtain an optimal solution to the given problem. In addition, it may influence the quality of the solution obtained. In the machine learning field, feature selection is an important process to attaining a good performance model; Genetic algorithm has been utilized for this purpose by scientists. However, the characteristics of Genetic algorithm, namely random initial population generation from a vector of feature elements, may influence solution and execution time. In this paper, the use of a statistical algorithm has been introduced (Chi2) for feature relevant checks where p-values of conditional independence were considered. Features with low p-values were discarded and subject relevant subset of features to Genetic Algorithm. This is to gain a level of certainty of the fitness of features randomly selected. An ensembled-based learning model for Hepatitis has been developed for Hepatitis C stage classification. 1385 samples were used using Egyptian-dataset obtained from UCI repository. The comparative evaluation confirms decreased in execution time and an increase in model performance accuracy from 56% to 63%.

• Journal of the Institute of Electronics and Information Engineers
• /
• v.51 no.5
• /
• pp.44-50
• /
• 2014

Chaos communication system is one of the security algorithms that is applied to improve the security. Chaos signal is non-linear, and it is generated randomly according to the initial conditions. Also, chaos communication system has characteristics such as non-periodic, wide-band, non-predictability of signals and easy implementation. So, security of chaos communication system is superior, and it has low interception probability and good anti-jamming characteristic. However, BER performance is worse than digital communication system, because it has many self interference signal in case of CDSK system. To improve these disadvantages, we analyze the PDF trend which can improve the BER performance in existing study, and we proposed a chaos map. And, proposed chaos map was defined as the 'Boss map'. Generally, BER performance is changed according to initial values, parameters and spreading factors. Therefore, in this paper, we will introduce PDF trends which can improve the BER performance, and will describe about Boss map. Also, characteristics of Boss map is analyzed by evaluating the BER performance of Boss map according to initial values, parameters and spreading factors. As a result, while maintaining the similar BER performance, initial value of Boss map can be selected from 0 to 1.2, and BER performance is best when parameter alpha is 2.5. Also, BER performance is best when spreading factor is 50.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.24 no.1
• /
• pp.83-88
• /
• 2024

In this research, we propose a Semi-Supervised learning based railroad surface defect detection method. The Resnet50 model, pretrained on ImageNet, was employed for the training. Data without labels are randomly selected, and then labeled to train the ResNet50 model. The trained model is used to predict the results of the remaining unlabeled training data. The predicted values exceeding a certain threshold are selected, sorted in descending order, and added to the training data. Pseudo-labeling is performed based on the class with the highest probability during this process. An experiment was conducted to assess the overall class classification performance based on the initial number of labeled data. The results showed an accuracy of 98% at best with less than 10% labeled training data compared to the overall training data.

• Asian-Australasian Journal of Animal Sciences
• /
• v.33 no.10
• /
• pp.1624-1632
• /
• 2020

Objective: A total of three hundred unsexed ducks were utilized to estimate net energy requirements of maintenance (NE_m) and weight gain (NE_g) for 2 to 3-week-old Cherry Valley ducks and to establish a model equation to predict NE requirements using the factorial method. Methods: To determine the apparent metabolizable energy (AME) of the diet, fifty 7-day-old ducks at approximately equal body weights (BWs) were randomly assigned into five groups that were fed at different levels (ad libitum, 85%, 75%, 65%, and 55% of ad libitum intake), and the endogenous acid-insoluble ash as indigestible marker. The two hundred and fifty 7-day-old ducks were used for a comparative slaughter experiment. At the beginning of the experiment, ten ducks were sacrificed to determine the initial body composition and energy content. The remaining ducks were randomly assigned into five groups (same as metabolic experiment). Ducks of the ad libitum group were slaughtered at 14 and 21-day-old. At the end of the experiment, two ducks were selected from each replicate and slaughtered to determine the body composition and energy content. Results: The results of the metabolizable experiment showed AME values of 13.43 to 13.77 MJ/kg for ducks at different feed intakes. The results of the comparative slaughter experiment showed the NE_m value for 2 to 3-week-old Cherry Valley ducks was 549.54 kJ/kg of BW^0.75/d, and the NE_g value was 10.41 kJ/g. The deposition efficiency values of fat (K_f) and crude protein (K_p) were 0.96 and 0.60, respectively, and the values of efficiency of energy utilization (K_g) and maintenance efficiency (K_m) were 0.75 and 0.88, respectively. Conclusion: The equation for the prediction of NE requirements for 2 to 3-week-old Cherry Valley ducks was the following: NE = 549.54 BW^0.75+10.41 ΔW, where ΔW is the weight gain (g).

• Korean Journal of Veterinary Research
• /
• v.27 no.2
• /
• pp.321-330
• /
• 1987

In the endemic area of bovine piroplasmosis in Chonbuk Province, total of 486 heads of Holstein cow selected from 14 farms randomly were examined the infection status of piroplasma and their blood values. The results obtained were summarized as follows: 1. The etiological agent was only Theileria sergenti and none of the Babesia sp. infection. 2. All of the 486 heads of examined cow were infected with Theileria sergenti (100%). 3. In the initial examination of 11 farms in early summer, the means of blood values were as follows; RBC:$526{\pm}84(10^4/mm^3$), Ht:$27.9{\pm}2.8%$, SP:$7.5{\pm}0.6g/dl$, WBC: $11586{\pm}354/mm^3$, and Fibrinogen: $578.5{\pm}164.1mg/dl$. 4. In the second examinations in slimmer season, the blood values of examined cows had little varieties. But in the winter season, cows housed in the barn, the blood values were greatly increased; RBC: $601{\pm}77(10^4/mm^3)$, (F-value: 37.186**), Ht: $30.5{\pm}1.3%$(F-value: 15.626**) Hb: $12.2{\pm}1.3g/dl$(F-value: 5.899**), SP: $7.4{\pm}0.6g/dl$(F-value: 5.05**). 5. To determine the piroplasma infection in the many herds of cow, the measurement of Ht, Hb, and SP values etc. were more effective and helpful to find the weak cow because the examined process are not only to be easily carry out many samples but also to be done very simple and rapidly. 6. The indications of therapeutic standard values were stablished as follows; the less than 20% in the Ht value, below the 10g/dl in the Hb value, and less than 6g/dl in SP value. When the one among the therapeutic standard values, Ht, Hb, and SP was below the therapeutic standard value, the cow was treated with nutriments. When more than two among them were below the therapeutic standard, the cow was treated with Berenil, at once. In this way, there was no one head which represented clinical signs of piroplasmosis among 486 heads of cow at 14 farms during the year of 1986. 7. In a Korean native cattle infected with Thieleria sergenti seriously, the recovery of blood values was very prolonged by the treatment of Berenil only. But by the transfusion of 300ml blood collected from normal cow, the blood values were recovered rapidly.

• Asian-Australasian Journal of Animal Sciences
• /
• v.23 no.12
• /
• pp.1645-1656
• /
• 2010

This study was conducted to investigate the effects of strain on broiler performance, and age at slaughter and postchilling (PC) aging time on meat quality traits. A total of 500 one-day-old chicks (250 Hubbard classic and 250 Lohman) were reared under commercial conditions. Half of the broiler birds from each strain were slaughtered at 32 days and the other half at 42 days old. At each processing day, 168 carcasses were randomly selected (84 Hubbard and 84 Lohman) and divided into groups of 28 carcasses within each strain, and aged for 0, 4 and 24 h after chilling. Average weekly body weight was comparable between strains. Feed conversion ratio was higher (p<0.05) for the Hubbard strain during the second and third week of age. Initial carcass pH was significantly (p<0.05) affected by age where younger birds (32-d-old) had lower pH values than older (41-d-old) birds. Breast temperature was higher (p<0.001) for Lohman than Hubbard at 0, 2 and 4 h of PC. Younger birds had a lower breast temperature (p<0.001) at all measured times of PC. Thaw loss, cook loss and water holding capacity were not significantly affected by strain, age or aging time. Lohman strain had more tender meat (p<0.05) than Hubbard strain, and tenderness was improved with the increase of broiler age and aging time. Meats from Hubbard were lighter and less red than those from Lohman strain where younger birds had darker color. In conclusion, strain, age at slaughter and PC aging duration are critical to breast meat quality characteristics, and 4 h of aging are required before deboning in order to obtain more tender fillets.

• Journal of Periodontal and Implant Science
• /
• v.24 no.1
• /
• pp.120-130
• /
• 1994

The purpose of the present investigation was to compare the effectiveness of porous hydroxyapatite (PHA) and coralline based porous calcium carbonate(PCC) as implant materials in human periodontal osseous defects. 10 adult patients having periodontitis and 2 similar angular osseous defects ${\ge}$5mm as verified by radiographic analysis and clinical probing depth ${\ge}$4mm were selected. The measurements were recorded just before surgery and after 6 month. Clinical parameters used in this study included gingival recession, pocket depth, probing attachment level, Sulcus Bleeding Index, Plaque Index, tooth mobility and bone defect depth measurements. After initial therapy, patients were treated with mucoperiosteal flap surgery. The contralateral bony defects in each patient randomly assigned to either bone graft material, one with PHA(Interpore 200) and the other with coralline based calcium carbonate(Biocoral). After 6 month both groups showed statistically significant reduction of pocket depth, Sulcus Bleeding Index, Plaque Index and significant improvement in probing attachment level. No statistically significant differences were found between the groups. There were 3.0mm or 68% of bone repair with PHA and 3.1mm, 61% with PCC. These values were likewise not significantly different. The data and clinical impression strongly suggest that both PHA and PCC are alloplastic implants with clinically apparent acceptance by the soft and hard tissue and that they can be used as bone graft materials successfully.