민첩한 활성함수를 이용한 합성곱 신경망의 성능 향상

공나영,고영민,고선우 한국정보처리학회 2020 정보처리학회논문지. 소프트웨어 및 데이터 공학 Vol.9 No.7

The convolutional neural network is composed of convolutional layers and fully connected layers. The nonlinear activation function is used in each layer of the convolutional layer and the fully connected layer. The activation function being used in a neural network is a function that simulates the method of transmitting information in a neuron that can transmit a signal and not send a signal if the input signal is above a certain criterion when transmitting a signal between neurons. The conventional activation function does not have a relationship with the loss function, so the process of finding the optimal solution is slow. In order to improve this, an agile activation function that generalizes the activation function is proposed. The agile activation function can improve the performance of the deep neural network in a way that selects the optimal agile parameter through the learning process using the primary differential coefficient of the loss function for the agile parameter in the backpropagation process. Through the MNIST classification problem, we have identified that agile activation functions have superior performance over conventional activation functions. 합성곱 신경망은 합성곱층과 완전연결층으로 구성되어 있다. 합성곱층과 완전연결층의 각 층에서는 비선형 활성함수를 사용하고 있다. 활성함수는 뉴런 간에 신호를 전달할 때 입력신호가 일정 기준 이상이면 신호를 전달하고 기준에 도달하지 못하면 신호를 보내지 않을 수 있는 뉴런의 정보전달 방법을 모사하는 함수이다. 기존의 활성함수는 손실함수와 관계성을 가지고 있지 않아 최적해를 찾아가는 과정이 늦어지는 점을 개선하기 위해 활성함수를 일반화한 민첩한 활성함수를 제안하였다. 민첩한 활성함수의 매개변수는 역전파 과정에서, 매개변수에 대한 손실함수의 1차 미분계수를 이용한 학습과정을 통해 최적의 매개변수를 선택하는 방법으로 손실함수를 감소시킴으로써 심층신경망의 성능을 향상시킬 수 있다. MNIST 분류문제를 통하여 민첩한 활성함수가 기존의 활성함수에 비해 우월한 성능을 가짐을 확인하였다.

TMS 기법을 이용한 학습자 모델링의 설계

공나영(Na Young Kong),박승수(Seung Soo Park) 한국정보과학회 1994 한국정보과학회 학술발표논문집 Vol.21 No.1

지능형 교수 시스템을 구성하는데 있어 학습자에 대한 완전한 정보가 불가능한 상태에서 학습자를 평가하기 위해서는 보다 효과적인 모델링이 요구된다. 본 연구에서는 믿음의 변환 및 디폴트 규칙의 표현에 적합한 지식 표현 방법인 TMS와 이의 변형인 ATMS를 학습자 모델링 설계에 적용시키는 방법을 제시하고 그 결과를 분석해 보고자 한다.

파라메트릭 활성함수를 이용한 심층신경망의 성능향상 방법

공나영(Nayoung Kong),고선우(Sunwoo Ko) 한국콘텐츠학회 2021 한국콘텐츠학회논문지 Vol.21 No.3

심층신경망은 임의의 함수를 근사화하는 방법으로 선형모델로 근사화한 후에 비선형 활성함수를 이용하여 추가적 근사화를 반복하는 근사화 방법이다. 이 과정에서 근사화의 성능 평가 방법은 손실함수를 이용한다. 기존 심층학습방법에서는 선형근사화 과정에서 손실함수를 고려한 근사화를 실행하고 있지만 활성함수를 사용하는 비선형 근사화 단계에서는 손실함수의 감소와 관계가 없는 비선형변환을 사용하고 있다. 본 연구에서는 기존의 활성함수에 활성함수의 크기를 변화시킬 수 있는 크기 파라메터와 활성함수의 위치를 변화시킬 수 있는 위치 파라미터를 도입한 파라메트릭 활성함수를 제안한다. 파라메트릭 활성함수를 도입함으로써 활성함 수를 이용한 비선형 근사화의 성능을 개선시킬 수 있다. 각 은닉층에서 크기와 위치 파라미터들은 역전파 과정에서 파라미터들에 대한 손실함수의 1차 미분계수를 이용한 학습과정을 통해 손실함수 값을 최소화시키는 파라미터를 결정함으로써 심층신경망의 성능을 향상시킬 수 있다. MNIST 분류 문제와 XOR 문제를 통하여 파라메트릭 활성함수가 기존의 활성함수에 비해 우월한 성능을 가짐을 확인하였다. Deep neural networks are an approximation method that approximates an arbitrary function to a linear model and then repeats additional approximation using a nonlinear active function. In this process, the method of evaluating the performance of approximation uses the loss function. Existing in-depth learning methods implement approximation that takes into account loss functions in the linear approximation process, but non-linear approximation phases that use active functions use non-linear transformation that is not related to reduction of loss functions of loss. This study proposes parametric activation functions that introduce scale parameters that can change the scale of activation functions and location parameters that can change the location of activation functions. By introducing parametric activation functions based on scale and location parameters, the performance of nonlinear approximation using activation functions can be improved. The scale and location parameters in each hidden layer can improve the performance of the deep neural network by determining parameters that minimize the loss function value through the learning process using the primary differential coefficient of the loss function for the parameters in the backpropagation. Through MNIST classification problems and XOR problems, parametric activation functions have been found to have superior performance over existing activation functions.

대구 소재 금연 시도 학생들의 특성에 대한 탐색적 연구

공나영(Nayeong Kong),김희진(Hee Jin Kim),서완석(Wan Seok Seo) 대한생물치료정신의학회 2021 생물치료정신의학 Vol.27 No.1

Objectives:Adolescents cigarette smoking is a major public health concern, because it is highly associated with physical and mental development of adolescents. This study examines the psychosocial variables affecting smoking behaviors and smoking cessation plan. Methods:A total of 74 (35 smokers and 40 non-smokers) students in Daegu metropolitan city were included in this study. 35 smokers visited smoking cessation center between September 2019 and November 2019. Subjects were asked to complete a self-reported questionnaire evaluating smoking status, self-esteem, parent-adolescent communication, parent and peer attachment, and peer relationship quality. The independent t-test was conducted to determine the difference in psychological variables between smoker and non-smoker. Additionally, logistic regression analysis was done to identify the factors that affect smoking cessation plan and willingness. Results:The majority of adolescent smoker was male (94.3%) and middle school students (82.9%). Twenty-four smokers (68.5%) started smoking at sixth and seventh grade. The smoking amount and smoking days were related to plan and willingness of smoking cessation in logistic regression analysis and the number of friends who smoke was related to smoking cessation plan. The openness of communication in smoking students was significantly higher than that of non-smoking students. Conclusion:Our findings indicate that the environmental factors such as peer or parental smoking is important in adolescents smoking. In addition to individual factors such as smoking amount and days, openness of communication related with impulsivity and susceptibility in peer pressure can affect plan or willingness of smoking cessation.

Effectiveness and Tolerability of Combination Pharmacotherapy With Stimulant and Non-Stimulant in Children With Attention Deficit Hyperactivity Disorder

박형남,공나영,김희철,김양태,정성원,이호준 대한소아청소년 정신의학회 2024 소아청소년정신의학 Vol.35 No.1

Objectives: This study aimed to investigate the effectiveness and safety of combining psychostimulants and nonstimulants for patients under treatment for attention deficit hyperactivity disorder (ADHD). Methods: The study included 96 patients aged 6–12 years who were diagnosed with ADHD, among whom 34 received combination pharmacotherapy, 32 received methylphenidate monotherapy, and 30 received atomoxetine monotherapy. Statistical analysis was conducted to compare treatment and adverse effects among groups and to analyze changes before and after combination pharmacotherapy. The difference between combination pharmacotherapy and monotherapy was investigated. Logistic regression analysis was used to identify the predictors of combination pharmacotherapy. Results: No significant differences were observed between the groups in terms of age or pretreatment scores. The most common adverse effect experienced by 32% of patients in the combination pharmacotherapy group was decreased appetite. Clinical global impression- severity score decreased significantly after combination pharmacotherapy. All three groups showed significant clinical global impression- severity score improvements over time, with no significant differences among them. The predictive factors for combination pharmacotherapy included the Child Behavior Checklist total score internalizing subscale. Conclusion: Combination pharmacotherapy with methylphenidate and atomoxetine is a relatively effective and safe option for patients with ADHD who do not respond to monotherapy.

건강보험 청구자료를 이용한 전립선염 환자의 의료 이용 및 항생제 처방 현황

이보람,최윤정,최연송,공나영,최민선,Lee, Boram,Choi, Yoon Jung,Choi, Younsong,Kong, Nayoung,Choi, Minsun 한국임상약학회 2018 한국임상약학회지 Vol.28 No.2

Background: Prostatitis, one of the most common diseases of the prostate, is a complex disease with various clinical features. This study aims to analyze the utilization and prescribing patterns of antibiotics in Korean patients with prostatitis between 2008 and 2015. Methods: We used the National Health Insurance Database complied from the Health Insurance Review and Assessment Service (HIRA). The outcomes included the number of claims, number of patients, medical cost, and length of stay for each year. In addition, the prescribing patterns of antibiotics, including fluoroquinolone, and low-dose use of ciprofloxacin and levofloxacin were investigated. Results: The total number of patients and medical cost increased by 9.5% and 51.7% from 2008 to 2015, respectively. Most prostatitis patients were classified as chronic prostatitis patients. The prescribing proportion of antibiotics for chronic prostatitis outpatients decreased from 71.0% to 66.9% from 2008 to 2015, and fluoroquinolone accounted for more than half of the total antibiotics. Over 80% of prescription of levofloxacin and ciprofloxacin was identified to be for low-dose use. Conclusion: Most of the patients with prostatitis experienced pain relief and condition improvement after antibiotic treatment; however, chronic prostatitis and chronic pelvic pain syndrome recur easily. Therefore, active disease management and further studies are needed to enhance our understanding of effective treatment for prostatitis.

전주대학교 사범대학 입학사정관제 인·적성 평가의 타당성 및 신뢰성 검증

하정 ( Jung Ha ),최희진 ( Hee Jin Choi ),공나영 ( Na Young Kong ) 전주대학교 사회과학연구소( 구 사회과학종합연구소) 2013 社會科學論叢 Vol.29 No.1

본 연구는 사범대학 2012학년도 입학사정관제 전형 중 수퍼스타전형과 지역인재균형선 발전형의 인·적성 평가의 타당성과 전형절차의 신뢰성을 검증하고자 하였다. 본 연구의 대상은 J대학교 사범대학 입학사정관전형 중 2012년 수퍼스타전형과 지역인재균형선발전형의 교직 인·적성평가의 서류평가와 면접평가의 내용, 전형 절차들이다. 이를 위해 각 평가영역을 인·적성으로 분류한 후 전문가 평정을 받아 평가내용의 타당성을 분석하였다. 연구 결과 내용적합도 부분에서 두 전형의 서류평가와 면접평가의 각 평가요소들의 평균이 4점 내외를 나타내어 긍정적인 수치를 나타냈다. 이는 내용 적합도가 높음을 의미한다. 반면 리더십과 사회기여활동 영역에서는 타당도의 기준치에 미치지 못하여 이 영역에서는 적절한 평가가 어렵다는 의견이 도출되었다. 따라서 이 두 영역에 대한 평가요소와 그 평가기준과 방식에 대한 점검과 수정이 요청된다. 전형절차의 신뢰성에 대한 평정 결과, 전형절차의 체계적인 일관성, 공정성, 그리고 입학사정관의 전문성 3가지 지표 모두 평균과 타당도가 기준 점수 이상으로 나타나 전형절차가 신뢰롭게 운영되고 있음이 밝혀졌다. 본 연구 결과는 차년도 사범대학 입학사정관 전형계획을 개선, 보완하는데 기초자료로 활용될 수 있으며 추후 관련 연구의 활성화에 기여할 것으로 기대된다. The purpose of this study is to verify the validity and reliability of the personality and aptitude evaluation of both the Superstar Program and the Regionally Balanced Admission Program of J University’s Department of Education’s Admission Officer System, 2012. This study mainly focuses on the content and the admission process of the document and interview evaluations of those two programs. Each evaluation item was divided into either an aptitude or personality section, then specialists graded them in order to evaluate the validity of the programs. The result of this study shows that both programs have a positive aspect regarding the average, which was 4, from each item of the document and interview evaluations. On the other hand, neither program is suitable to evaluate applicants’ leadership or social contribution aspects. Their evaluation items, criterion and methods require further examination and correction. In addition, the reliability of the whole procedure hit higher scores than the average in their consistency, fairness and expertise. This study provides the basis for the following development of the Department of Education’s Admission Programs.