• Convergence Security Journal
• /
• v.18 no.3
• /
• pp.53-59
• /
• 2018

In this paper, we propose a hierarchical image classification scheme for efficient object image classification. In the non-hierarchical image classification, which classifies the existing whole images at one time, it showed that objects with relatively similar shapes are not recognized efficiently. Therefore, in this paper, we introduce the image classification method in the hierarchical structure which attempts to classify object images hierarchically. Also, we introduce to the efficient class structure and algorithms considering the scalability that can occur when a deep learning image classification is applied to an actual system. Such a scheme makes it possible to classify images with a higher degree of confidence in object images having relatively similar shapes.

• Journal of the Institute of Electronics Engineers of Korea SC
• /
• v.48 no.5
• /
• pp.7-12
• /
• 2011

In this papwer, we propose a new feature extraction method, named as Class-augmented Kernel Principal Component Analysis (CA-KPCA), which can extract nonlinear features for classification. Among the subspace method that was being widely used for feature extraction, Class-augmented Principal Component Analysis (CA-PCA) is a recently one that can extract features for a accurate classification without computational difficulties of other methods such as Linear Discriminant Analysis (LDA). However, the features extracted by CA-PCA is still restricted to be in a linear subspace of the original data space, which limites the use of this method for various problems requiring nonlinear features. To resolve this limitation, we apply a kernel trick to develop a new version of CA-PCA to extract nonlinear features, and evaluate its performance by experiments using data sets in the UCI Machine Learning Repository.

• Journal of Chest Surgery
• /
• v.28 no.9
• /
• pp.822-828
• /
• 1995

This paper reports 15 native valve endocarditis cases had surgical operation in the past 10 years at the department of Cardiovascular and Thoracic Surgery, Chonbuk National University Hospital. In this study, 10 cases out of 15 were in class I or II by the New York Heart Association functional classification. None of the cases had a history of taking addictive drugs. Five cases were congenital heart disease, three cases were rheumatic heart disease and two cases were degenerative heart disease. Thus 10 cases had the underlying disease. All cases had antibiotics treatment for 3 to 6 weeks before operation. In the culture test, only four cases were positive in the blood culture and one case was positive in the excised valve culture. Organisms on blood and valve culture were Streptococcus epidermis, Streptococcus viridans, Staphylococcus aureus and Staphylococcus epidermidis. In the 10 cases without ventricular septal defect, the aortic valve was involved in four, mitral in four, both in two and involved valves in the 5 cases with ventricular septal defect were tricuspid in three, pulmonic in two. Eight cases had operation because they showed moderate congestive heart failure due to valvular insufficiency and vegetation with or without embolism. Seven cases had operation because they showed persistent or progressive congestive heart failure and/or uncontrolled infection. Five cases with ventricular septal defect underwent the closure of ventricular septal defect, vegetectomy and leaflet excision of the affected valves without valve replacement. In the cases without ventricular septal defect, the affected valves were replaced with St. Jude mechanical prosthesis. Postoperative complications were recurrent endocarditis in two, embolism in one, allergic vasculitis in two, spleen rupture in one and postpericardiotomy syndrome in one. At the first postoperative day, one case died of cerebral embolism. At the 11th postoperative month, one case died of recurrent endocarditis and paravalvular leakage in spite of a couple of aortic valve replacement. In the survived cases[13 cases in this study , all cases but one became class I or II by the New York Heart Association functional classification.

• Journal of the Institute of Electronics Engineers of Korea SP
• /
• v.40 no.3
• /
• pp.203-216
• /
• 2003

In this paper a new feature extraction and target classification method is proposed for the recognition part of FLIR(Forwar Looking Infrared)-image-based ATR system. Proposed feature extraction method is "cluster(=set of classes)-based"version of previous fisherfaces method that is known by its robustness to illumination changes in face recognition. Expecially introduced class clustering and cluster-based projection method maximizes the performance of fisherfaces method. Proposed target image classification method is based on the mixture of experts model which consists of RBF-type experts and MLP-type gating networks. Mixture of experts model is well-suited with ATR system because it should recognizee various targets in complexed feature space by variously mixed conditions. In proposed classification method, one expert takes charge of one cluster and the separated structure with experts reduces the complexity of feature space and achieves more accurate local discrimination between classes. Proposed feature extraction and classification method showed distinguished performances in recognition test with customized. FLIR-vehicle-image database. Expecially robustness to pixelwise sensor noise and un-wanted intensity variations was verified by simulation.

• The korean journal of orthodontics
• /
• v.21 no.2 s.34
• /
• pp.457-468
• /
• 1991

This investigation was designed to categorize Angle's class III malocclusion groups through analyzing horizontal and vertical components of craniofacial skeleton in Angle's class III malocclusion. The material selected for this study consisted in standard lateral cephalogram of one hundred and fifty five children, seventy four boys and eighty one girls, aged 6 through 15 years, having Angle's class III malocclusion. On the basis of findings of this study, the following results were obtained. 1. In horizontal skeletal classifications,16 groups were classified FMN-A-B, SE-FMN-A according to the Ba-SE-Me, Ba-SE/R. 2. The sequences that have relatively high frequency are as follow; a) Horizontal Group l b) Horizontal Group 3 c) Horizontal Group 5 d) Horizontal Group 9 e) Horizontal Group 4 3. In vertical skeletal classification, 8 groups were classified according to the PMV/PP, PMV/OP, PMV/MP. 4. The sequences that relatively high are as follows; a) Vertical Group 1 b) Vertical Group 3 c) Vertical Group 4 d) Vertical Group 8.

• Journal of Chest Surgery
• /
• v.28 no.9
• /
• pp.811-816
• /
• 1995

Repairs of atrial septal defect utilizing cardiopulmonary bypass were performed in 50 adults, ranging age from 16 to 53 years, since April 1986 up to October 1994. They occupied 38.8% of all adult congenital heart disease operated in the same period. Preoperatively, 16 patients were functional class II, 12 patients class III and 4 patients class IV[New York Heart Association Classification , respectively. Five patients combined with atrial fibrillation and the remainders revealed regular sinus rhythm. Cardiac catheterizations were performed in 43 out of 50 patients, and revealed a systolic pulmonary arterial pressure in excess of 51 mmHg in 4 patients but none had reversed shunt. 40 patients[80% were repaired with patch closure and remainings were repaired with direct closure. 49 patients were followed up for 2 months up to 102 months[average 55 months . A comparison of the preoperative and postoperative functional class demonstrated a mean decrease of one NYHA functional level[2.5$\pm$0.63 to 1.4$\pm$0.56 . There was no operative mortality. One patient died during the follow-up period and the death was unrelated to heart disease. Operative treatment is indicated for repair of atrial septal defect with left to right shunt in the adult patient and a considerable clinical improvement can be anticipitated with low mortality.

• Journal of Chest Surgery
• /
• v.26 no.7
• /
• pp.521-525
• /
• 1993

Sixty-three patients who underwent open mitral commissurotomy at Sejong General Hospital during last 10 years from the August, 1983 to June, 1993 were reviewed There were fifty-one women and twelve men, and the mean age was 34.2 years. According to the NYHA classification, the distribution of patient preoperatively was as follows: class I, 5 patients ; class II, 26 patients ; class III, 30 patients ; class IV, 2 patients, and mean duration of symptome was 4.5 years. The mean mitral valvular area measured with echocardiogram preoperatively was 0.92cm2. All patients underwent open mitral commissurotomy and 41 patients required additional procedures for relief of subvalvular stenosis or other valvular disease. There were no operative death. 63 patients had three different types of mitral stenosis : type I, mobile cusps without subvalvular change [21 patients] ; typeII, thickened cusps with subvalvular change [34 patients] ; type III, rigid cusps with severe subvalvular change [8 patients]. The valvular calcification was seen in 11 patients [17%] and 15 patients [24%] had left atrial thrombus. The duration of follow-up was from 1 month to 168 months [mean, 39.6 months] and there were no late death. Six patients required reoperation and one patient had embolic episode. Conclusively,the open mitral commissurotomy represents a safe surgical precedure for treating mitral stenosis, allowing a complete removal of atrial thrombosis if present, and even when associated with subvalvular changes.

• Korean Journal of Clinical Pharmacy
• /
• v.25 no.2
• /
• pp.120-129
• /
• 2015

Objectives: Drug utilization review program in Korea has provided 'drug combinations to avoid (DCA)' alerts to physicians and pharmacists to prevent potential adverse drug events or inappropriate drug use. Seven hundred and six DCA pairs have been announced officially by the Ministry of Food and Drug Safety (MFDS) by March, 2015. Some DCA pairs could be grouped based on the drug interaction mechanism and its consequences. This study aimed to investigate the drug-drug interaction (DDI) pairs, which may be potential DCAs, generated by the drug class-drug class interaction method. Methods: Eleven additive/synergistic and one antagonistic drug class-drug class interaction groups were identified. By combining drugs of two interacting drug class groups, numerous DDI pairs were made. The status and severity of DDI pairs were examined using Lexicomp and Micromedex. Also, the DCA listing rate was calculated. Results: Among 258 DDI pairs generated by the drug class-drug class interaction method, only 142 pairs were identified as official DCA pairs by the MFDS. One hundred and four pairs were identified as potential DCA pairs to be listed. QT prolonging agents-QT prolonging agents, triptans-ergot alkaloids, tricyclic antidepressants-monoamine oxidase inhibitors, and dopamine agonists-dopamine antagonists were identified as drug class-drug class interaction groups which have less than 50 % DCA listing rate. Conclusion: To improve the clinicians' adaptability to DCA alerts, the list of DCA pairs needs to be continuously updated.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.12 no.8
• /
• pp.3725-3748
• /
• 2018

Text data distribution is often imbalanced. Imbalanced data is one of the challenges in text classification, as it leads to the loss of performance of classifiers. Many studies have been conducted so far in this regard. The proposed solutions are divided into several general categories, include sampling-based and algorithm-based methods. In recent studies, feature selection has also been considered as one of the solutions for the imbalance problem. In this paper, a novel one-sided feature selection known as probabilistic feature selection (PFS) was presented for imbalanced text classification. The PFS is a probabilistic method that is calculated using feature distribution. Compared to the similar methods, the PFS has more parameters. In order to evaluate the performance of the proposed method, the feature selection methods including Gini, MI, FAST and DFS were implemented. To assess the proposed method, the decision tree classifications such as C4.5 and Naive Bayes were used. The results of tests on Reuters-21875 and WebKB figures per F-measure suggested that the proposed feature selection has significantly improved the performance of the classifiers.

• Journal of the Korea Institute of Information Security & Cryptology
• /
• v.31 no.6
• /
• pp.1227-1236
• /
• 2021

Adversarial attack, one of the attacks on deep learning classification model, is attack that add indistinguishable perturbations to input data and cause deep learning classification model to misclassify the input data. There are various adversarial attack algorithms. Accordingly, many studies have been conducted to detect adversarial attack but few studies have been conducted to classify what adversarial attack algorithms to generate adversarial input. if adversarial attacks can be classified, more robust deep learning classification model can be established by analyzing differences between attacks. In this paper, we proposed a model that detects and classifies adversarial attacks by constructing a random forest classification model with input features extracted from a target deep learning model. In feature extraction, feature is extracted from a output value of hidden layer based on class predicted by the target deep learning model. Through Experiments the model proposed has shown 3.02% accuracy on clean data, 0.80% accuracy on adversarial data higher than the result of pre-existing studies and classify new adversarial attack that was not classified in pre-existing studies.