• ETRI Journal
• /
• v.46 no.3
• /
• pp.513-525
• /
• 2024

Deep neural networks trained on labeled medical data face major challenges owing to the economic costs of data acquisition through expensive medical imaging devices, expert labor for data annotation, and large datasets to achieve optimal model performance. The heterogeneity of diseases, such as Alzheimer's disease, further complicates deep learning because the test cases may substantially differ from the training data, possibly increasing the rate of false positives. We propose a reconstruction-based self-supervised anomaly detection model to overcome these challenges. It has a dual-subnetwork encoder that enhances feature encoding augmented by skip connections to the decoder for improving the gradient flow. The novel encoder captures local and global features to improve image reconstruction. In addition, we introduce an entropy-based image conversion method. Extensive evaluations show that the proposed model outperforms benchmark models in anomaly detection and classification using an encoder. The supervised and unsupervised models show improved performances when trained with data preprocessed using the proposed image conversion method.

• Journal of Astronomy and Space Sciences
• /
• v.23 no.2
• /
• pp.117-126
• /
• 2006

The Diurnal variation of galactic cosmic ray (GCR) flux intensity observed by the ground Neutron Monitor (NM) shows a sinusoidal pattern with the amplitude of $1{\sim}2%$ of daily mean. We carried out a statistical study on tendencies of the local times of GCR intensity daily maximum aad minimum. To test the influences of the solar activity and the location (cut-off rigidity) on the distribution in the local times of maximum and minimum GCR intensity, we have examined the data of 1996 (solar minimum) and 2000 (solar maximum) at the low-latitude Haleakala (latitude: 20.72 N, cut-off rigidity: 12.91 GeV) and the high-latitude Oulu (latitude: 65.05 N, cut-off rigidity: 0.81 GeV) NM stations. The most frequent local times of the GCR intensity daily maximum and minimum come later about $2{\sim}3$ hours in the solar activity maximum year 2000 than in the solar activity minimum you 1996. Oulu NM station whose cut-off rigidity is smaller has the most frequent local times of the GCR intensity maximum and minimum later by $2{\sim}3$ hours from those of Haleakala station. This feature is more evident at the solar maximum. The phase of the daily variation in GCR is dependent upon the interplanetary magnetic field varying with the solar activity and the cut-off rigidity varying with the geographic latitude.

• Korean journal of communication and information
• /
• v.46
• /
• pp.108-147
• /
• 2009

This study explores the news frames on the U. S. beef imports and candlelight vigils covered by the two national dailies such as ChosunIlbo and the Hankyoreh Shinmun; the KwangwonIlbo, a local daily. The news frames extracted based on the models of Iyengar(1987), Semetko & Valkenburg(2000) and other researchers are attribution of responsibility, economic sequences, protest against the authorities, national health and governmental public relations and so on. The result shows that the news reports are consisted of the straight news(75.9%), feature stories(11.7%) and editorials(6.3%). More specifically, there is a comparatively hight ratio of editorials(11.0%) for the ChosunIlbo, feature stories(20.9%) for the Hankyoreh, and the straight news(89.7%) for the KwangwonIlbo. In terms of the news frames stressed by the three dailies, the ChosunIlbo focuses and stresses on the national health(17.8%) and the attribution of responsibilities(10.6%). However, the Hankyoreh have a tendency to stress on the protest against the authorities(31.3%) and attribution of responsibilities(38.4%); the KwangwonIlbo, focuses on the protest against the authorities(38.4%) and the economic sequences(17.9%). Finally, in the case of the main characteristics of the dailies, the governmental public relations frame is found only on the ChosunIlbo that has a comparatively high ratio; the Hankyoreh also has a high ratio of the feature stories on the U. S. beef imports. Even thought the KwangwonIlbo has a high ratio of the economic sequence frame, the ratio of opinion pages, such as editorial and columns, the local newspaper has not spoken up for the potential economic crisis of the local Kwangwon province beef industry, mainly caused by the U. S. import beef.

• Korean Journal of Head & Neck Oncology
• /
• v.17 no.2
• /
• pp.205-209
• /
• 2001

Purpose: Most cases of primary hyperparathyroidism are due to parathyroid adenoma or parathyroid hyperplasia. Parathyroid carcinoma is a very rare cause of hyperparathyroidism. Although the diagnosis of parathyroid carcinoma is usually established by pathologic criteria especially of vascular or capsular invasion, some clinical and biochemical features differentiate it from benign forms of hyperparathyroidism. We under-took a retrospective study in 6 patients with parathyroid carcinoma, with the aim of conveying experience from management for this rare cause of hyperparathyroidism. Methods: Clinical symptoms, biochemical laboratory, radiologic, and intraoperative findings, local recurrence and distant metastasis were analyzed in 6 patients diagnosed pathologically as a parathyroid carcinoma after operation from 1992 to 2001. Results: Mean age was 50.2 years (33.0-60.0 years) and male to female ratio was 1:1. Neck mass was found in 5 patients, multiple bone pain in 3 patients and renal stone in 1 patient. One case has suffered from chronic renal failure for 19 years. Although preoperative laboratory evaluations showed the aspects of hyperparathyroidism in all cases, mean serum calcium level was 11.2mg/dl(10.5-12.1mg/dl), slightly elevated. Laboratory values after surgery were within the normal range in 5 cases. However, in one case with chronic renal failure, serum PTH levels, serially checked, were above the normal range. Any of imaging methods failed to suggest a parathyroid carcinoma preoperatively. Parathyroid adenoma was suspected in 3 cases, thyroid cancer in the other cases before surgery. The extent of resection was radical resection of parathyroid lesion with more than unilateral thyroid lobectomy and central compartment neck node dissection and in 2 cases, the resection of recurrent laryngeal nerve or strap muscles was added. During follow-up period, any local or systemic recurrence were not evident in all the cases. Conclusion: Although parathyroid carcinoma is a rare disease and its preoperative diagnosis, in our experience, could not easily be made, the understanding of characteristic clinical and biochemical feature could help diagnosis at first surgery. Radical resection without remaining residual tumor is most important for the management of the parathyroid cancer.

• Journal of the Institute of Electronics and Information Engineers
• /
• v.50 no.8
• /
• pp.187-195
• /
• 2013

Most of former studies for car license plate detection restrict the image acquisition environment. The aim of this research is to diminish the restrictions by proposing a new method of using SIFT and neural network. SIFT can be used in diverse situations with less restriction because it provides size- and rotation-invariance and large discriminating power. SIFT extracted from the license plate image is divided into the internal(inside class) and the external(outside class) ones and the classifier is trained using them. In the proposed method, by just putting the various types of license plates, the trained neural network classifier can process all of the types. Although the classification performance is not high, the inside class appears densely over the plate region and sparsely over the non-plate regions. These characteristics create a local feature map, from which we can identify the location with the global maximum value as a candidate of license plate region. We collected image database with much less restriction than the conventional researches. The experiment and evaluation were done using this database. In terms of classification accuracy of SIFT keypoints, the correct recognition rate was 97.1%. The precision rate was 62.0% and recall rate was 50.2%. In terms of license plate detection rate, the correct recognition rate was 98.6%.

• Korean Journal of Heritage: History & Science
• /
• v.45 no.4
• /
• pp.22-37
• /
• 2012

Storage facility of the pre-historic Korea is classified into the subterranean, ground and overground types. The rectangular-shaped subterranean Storage facilities have been identified in the Pungnaptoseong, the Gong mountain fortress site and the Gwanbukri site. Feature no. 196 at Gyeongdang District in Pungnaptoseong yielded a large quantity of glazed potteries, and a wooden storage at the Gwanbukri site contained a large amount of fruit seeds. These storage facilities might be functioned as the warehouse for the highest group rather than the storage for the emergency such as war and flooding and stipend of government officials. This article subdivided into "state storage" on the concept of the former that "royal storage" on the concept of the latter. If it look on the state storage at large, this include the royal storage too. But it subdivided to help article understand because Baekjae changed from state storage to royal storage by change and specialization of system after 4th. The reason why the diversification of storage pits was closely related to the unification of local polities and the concentration of political power in the state-level. Therefore, it might reflect the political circumstances the ruling elites attempted to heighten their authority in terms of the organizing tax collecting system. And divided the time of storage is confirmed separative storage pits in the suburbs of capital city. There is hight probability of top local polities or nation that have possessional a role. This is to cover on frequent war in the Three States. On the other hand, state storage is located around ancent road that linked castle gate that is divided into center and periphery depending on function and position of storage. Center is located royal storage focusing in the presumed royal palace that periphery is located state storage to provide service to the public. It is presumed that located with the government office.

• The Transactions of the Korea Information Processing Society
• /
• v.13 no.4
• /
• pp.199-207
• /
• 2024

Local reservoirs are crucial sources for agricultural water supply, necessitating stable water level management to prepare for extreme climate conditions such as droughts. Water level prediction is significantly influenced by local climate characteristics, such as localized rainfall, as well as seasonal factors including cropping times, making it essential to understand the correlation between input and output data as much as selecting an appropriate prediction model. In this study, extensive multivariate data from over 400 reservoirs in Jeollabuk-do from 1991 to 2022 was utilized to train and validate a water level prediction model that comprehensively reflects the complex hydrological and climatological environmental factors of each reservoir, and to analyze the impact of each input feature on the prediction performance of water levels. Instead of focusing on improvements in water level performance through neural network structures, the study adopts a basic Feedforward Neural Network composed of fully connected layers, batch normalization, dropout, and activation functions, focusing on the correlation between multivariate input data and prediction performance. Additionally, most existing studies only present short-term prediction performance on a daily basis, which is not suitable for practical environments that require medium to long-term predictions, such as 10 days or a month. Therefore, this study measured the water level prediction performance up to one month ahead through a recursive method that uses daily prediction values as the next input. The experiment identified performance changes according to the prediction period and analyzed the impact of each input feature on the overall performance based on an Ablation study.

• Journal of the Korean Society of Surveying, Geodesy, Photogrammetry and Cartography
• /
• v.24 no.1
• /
• pp.131-138
• /
• 2006

Recently, data gathering and algorithm developing are in progress for the GIS application using point feature. Several researches prove that verification of the spatial clustering and evaluation of inter-dependencies between event and control are possible. On the other hand, most of the point features as GIS data are gathered by indirect method, such as address geo-coding, rather than by direct method, such as field surveying. Futhermore, lots of statistics by administrative district based on the point features have no coordinates information of the points. In this study, calculating the K-index in GIS environment, to evaluate the effect of aggregation of raw data on K-index, K-indices estimated from raw data (parcel unit), topographically aggregated data (block unit), administratively aggregated data (administrative district unit) are compared and evaluated. As a result, point feature, highly clustered in local area, is largely distorted when aggregated administratively. But, the K-indices of topographically aggregated data is very similar to the K-indices of raw data.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.9 no.5
• /
• pp.163-168
• /
• 2009

Most image segmentation methods are to represent observed feature vectors at each pixel, which are assumed as appropriated probability models. These models can be used by statistical estimating or likelihood clustering algorithms of feature vectors. EM algorithms have some calculation problems of maximum likelihood for unknown parameters from incomplete data and maximum value in post probability distribution. First, the performance is dependent upon starting positions and likelihood functions are converged on local maximum values. To solve these problems, we mixed the Gausian function and histogram at all the level values at the image, which are proposed most suitable image segmentation methods. This proposed algoritms are confirmed to classify most edges clearly and variously, which are implemented to MFC programs.

• Journal of the Korea Society of Computer and Information
• /
• v.18 no.4
• /
• pp.19-26
• /
• 2013

Pseudo sample neural network (PSNN) is a variant of traditional neural network using pseudo samples to mitigate the local-optima-convergence problem when the size of training samples is small. PSNN can take advantage of the smoothed solution space through the use of pseudo samples. PSNN has a focus on the quantity problem in training, whereas, methods stressing the quality of training samples is presented in this paper to improve further the performance of PSNN. It is evident that typical samples and highly correlated features help in training. In this paper, therefore, kernel density estimation is used to select typical samples and correlation factor is introduced to select features, which can improve the performance of PSNN. Debris flow data set is used to demonstrate the usefulness of the proposed methods.