• 제목/요약/키워드: non-learning term

검색결과 86건 처리시간 0.028초

선형 활성화 함수를 이용한 개선된 퍼지 단층 퍼셉트론 (An Enhanced Fuzzy Single Layer Perceptron With Linear Activation Function)

  • 박충식;조재현;김광백
    • 한국정보통신학회논문지
    • /
    • 제11권7호
    • /
    • pp.1387-1393
    • /
    • 2007
  • 기존의 단층 퍼셉트론은 출력 노드가 선형 분리 가능한 패턴들만을 분류할 수 있고 XOR과 같은 비선형 문제에 대해서는 분류할 수 없는 단점이 있다. 퍼지 단층 퍼셉트론은 퍼지 소속 함수(Fuzzy Membership Function)를 적용하여 단층 구조로 XOR 문제와 같은 고전적인 문제를 개선하였다. 그러나 퍼지 단층 퍼셉트론은 기존의 단층 퍼셉트론과 마찬가지로 결정 경계선이 진동하는 경우가 생기며 초기 가중치의 범위와 학습률에 따라 수렴성이 매우 낮아지는 단점이 있다. 따라서 본 논문에서는 바이어스항을 도입하여 결정 경계선이 진동하는 것을 방지하여 수렴성을 개선시키고 선형 활성화 함수를 제안하고 학습률과 모멘텀 개념을 도입 한 개선된 델타규칙을 적용함으로써 학습 시간을 단축시키는 개선된 퍼지 단층 퍼셉트론 알고리즘을 제안한다. 제안된 방법과 퍼지 단층 퍼셉트론간의 학습 성능을 분석하기 위하여 인공 신경망에서 벤치마크로 사용되는 XOR 문제와 패턴 분류에 적용하여 Epoch 수와 수렴성을 비교한 결과, 제안된 방법이 기존의 퍼지 단층 퍼셉트론보다 학습 시간이 적게 소요되고 수렴성이 개선된 것을 확인하였다.

지자체 사이버 공간 안전을 위한 금융사기 탐지 텍스트 마이닝 방법 (Financial Fraud Detection using Text Mining Analysis against Municipal Cybercriminality)

  • 최석재;이중원;권오병
    • 지능정보연구
    • /
    • 제23권3호
    • /
    • pp.119-138
    • /
    • 2017
  • 최근 SNS는 개인의 의사소통뿐 아니라 마케팅의 중요한 채널로도 자리매김하고 있다. 그러나 사이버 범죄 역시 정보와 통신 기술의 발달에 따라 진화하여 불법 광고가 SNS에 다량으로 배포되고 있다. 그 결과 개인정보를 빼앗기거나 금전적인 손해가 빈번하게 일어난다. 본 연구에서는 SNS로 전달되는 홍보글인 비정형 데이터를 분석하여 어떤 글이 금융사기(예: 불법 대부업 및 불법 방문판매)와 관련된 글인지를 분석하는 방법론을 제안하였다. 불법 홍보글 학습 데이터를 만드는 과정과, 데이터의 특성을 고려하여 입력 데이터를 구성하는 방안, 그리고 판별 알고리즘의 선택과 추출할 정보 대상의 선정 등이 프레임워크의 주요 구성 요소이다. 본 연구의 방법은 실제로 모 지방자치단체의 금융사기 방지 프로그램의 파일럿 테스트에 활용되었으며, 실제 데이터를 가지고 분석한 결과 금융사기 글을 판정하는 정확도가 사람들에 의하여 판정하는 것이나 키워드 추출법(Term Frequency), MLE 등에 비하여 월등함을 검증하였다.

대학 장기현장실습 프로그램의 성과평가지표 개발 및 적용에 관한 연구 : 한국기술교육대학교의 사례 (A Study on the Development and Application of Performance Evaluation Criteria for Co-op Programs in Universities: Focused on the Case of KOREATECH)

  • 오창헌;엄기용
    • 경영과학
    • /
    • 제32권4호
    • /
    • pp.155-173
    • /
    • 2015
  • Koreatech has adopted a long-term co-op program called IPP (Industry Professional Practice) to address problems in higher education of Korea since 2012, but it was anticipated to face many difficulties in implementing the program due to lack of relevant experiences in Korea. In this regard, a performance evaluation scheme was urgently required to judge the effectiveness of the Co-op program and improve the operational efficiency at the same time. This study aimed to develop comprehensive performance evaluation criteria for the co-op programs on the basis of Kirkpatrick's four-level performance evaluation model for training programs, and apply it to a real Co-op operation to test its feasibility. For this purpose, thorough review on the training program evaluation literature and in-depth analyses of overseas cases of co-op performance evaluation were conducted. Then, a set of Co-op performance evaluation criteria was developed and applied to the Koreatech's Co-op operation in 2013. Two Co-op student questionnaire surveys were administered before and after Co-op terms to measure students' reactions to the Co-op program (Level 1) and changes in participating students' attitudes and competencies (Level 2). In addition, employment rates of Co-op participating and non-participating students were compared (Level 4). The analysis findings showed that participating students were quite satisfied with their jobs and companies, and the Co-op program was effective to a certain degree at enhancing non-major competencies and attitudes to occupation of students. Together, the employment rate of Co-op participants, particularly in small- and medium-sized companies, grew significantly in comparison with that of non-participants. In the last part, several directions for improving the effectiveness of the Co-op programs were discussed.

Preparation of Alzheimers Animal Model and Brain Dysfunction Induced by Continuous $\beta$-Amyloid Protein Infusion

  • Akio Itoh;Kiyofumi Yamada;Kim, Hyoung-Chun;Toshitaka Nabeshima
    • Toxicological Research
    • /
    • 제17권
    • /
    • pp.47-57
    • /
    • 2001
  • Alzheimer's disease (AD) is the most common cause of dementia in the elderly, and its pathology is characterized by the presence of numerous numbers of senile plaques and neurofibrillary tangles. Several genetic and transgenic studies have indicated that excess amount of $\beta$-amyloid protein (A$\beta$) is produced by mutations of $\beta$TEX>$\beta$-amyloid precursor protein and causes learning impairment. Moreover, $A\beta$ has a toxic effect on cultured nerve cells. To prepare AD model animals, we have examined continuous (2 weeks) infusion of $A\beta$ into the cerebral ventricle of rats. Continuous infusion of $A\beta$ induces learning impairment in water maze and passive avoidance tasks, and decreases choline acetyltransferase activity in the frontal cortex and hippocampus. Immunohistochemical analysis revealed diffuse depositions of $A\beta$ in the cerebral cortex and hippocampus around the ventricle. Furthermore, the nicotine-evoked release of acetylcholine and dopamine in the frontal cortex/hippocampus and striatum, respectively, is decreased in the $A\beta$-infused group. Perfusion of nicotine (50 $\mu\textrm{M}$) reduced the amplitude of electrically evoked population spikes in the CA1 pyramidal cells of the control group, but not in those of the $A\beta$-infused group, suggesting the impairment of nicotinic signaling in the $A\beta$-infused group. In fact, Kd, but not Bmax, values for [$^3H$] cytisine binding in the hippocampus significantly increased in the $A\beta$-infused rats. suggesting the decrease in affinity of nicotinic acetylcholine receptors. Long-term potentiation (LTP) induced by tetanic stimulations in CA1 pyramidal cells, which is thought to be an essential mechanism underlying learning and memory, was readily observed in the control group, whereas it was impaired in the $A\beta$-infused group. Taken together, these results suggest that $A\beta$ infusion impairs the signal transduction mechanisms via nicotinic acetylcholine receptors. This dysfunction may be responsible, at least in part, for the impairment of LTP induction and may lead to learning and memory impairment. We also found the reduction of glutathione- and Mn-superoxide dismutase-like immunoreactivity in the brains of $A\beta$-infused rats. Administration of antioxidants or nootropics alleviated learning and memory impairment induced by $A\beta$ infusion. We believe that investigation of currently available transgenic and non-transgenic animal models for AD will help to clarify the pathogenic mechanisms and allow assessment of new therapeutic strategies.

  • PDF

기계학습 모형 기반 진해만 용존산소농도 및 빈산소수괴 발생 예측 (Prediction in Dissolved Oxygen Concentration and Occurrence of Hypoxia Water Mass in Jinhae Bay Based on Machine Learning Model)

  • 박성식;김병국;김경회
    • 한국해안·해양공학회논문집
    • /
    • 제34권3호
    • /
    • pp.47-57
    • /
    • 2022
  • 본 연구에서는 진해만의 단일 정점 장기 모니터링 자료를 사용하여 LSTM 모형을 이용한 DO 농도 예측 및 결정 트리 모형을 이용한 빈산소수괴 발생 예측 연구를 수행하였다. LSTM을 이용한 DO 농도 예측 결과, Hidden node의 수가 증가할수록 모형의 복잡도가 증가하여 많은 Epoch을 요구하는 모습을 보였으며, 예측 시간 간격이 증가할수록 긴 Sequence length에서 높은 정확도를 보였다. 결정 트리를 이용한 빈산소수괴 발생 예측 결과, 30 day 예측에서 빈산소수괴 미발생 예측 정확도는 6 6 .1%로 발생 예측 정확도의 37.5%보다 상대적으로 높게 나타났다. 이는 결정 트리 모형이 DO 농도를 관측값보다 고평가하여 나타난 결과로 판단된다.

The evaluation of Spectral Vegetation Indices for Classification of Nutritional Deficiency in Rice Using Machine Learning Method

  • Jaekyeong Baek;Wan-Gyu Sang;Dongwon Kwon;Sungyul Chanag;Hyeojin Bak;Ho-young Ban;Jung-Il Cho
    • 한국작물학회:학술대회논문집
    • /
    • 한국작물학회 2022년도 추계학술대회
    • /
    • pp.88-88
    • /
    • 2022
  • Detection of stress responses in crops is important to diagnose crop growth and evaluate yield. Also, the multi-spectral sensor is effectively known to evaluate stress caused by nutrient and moisture in crops or biological agents such as weeds or diseases. Therefore, in this experiment, multispectral images were taken by an unmanned aerial vehicle(UAV) under field condition. The experiment was conducted in the long-term fertilizer field in the National Institute of Crop Science, and experiment area was divided into different status of NPK(Control, N-deficiency, P-deficiency, K-deficiency, Non-fertilizer). Total 11 vegetation indices were created with RGB and NIR reflectance values using python. Variations in nutrient content in plants affect the amount of light reflected or absorbed for each wavelength band. Therefore, the objective of this experiment was to evaluate vegetation indices derived from multispectral reflectance data as input into machine learning algorithm for the classification of nutritional deficiency in rice. RandomForest model was used as a representative ensemble model, and parameters were adjusted through hyperparameter tuning such as RandomSearchCV. As a result, training accuracy was 0.95 and test accuracy was 0.80, and IPCA, NDRE, and EVI were included in the top three indices for feature importance. Also, precision, recall, and f1-score, which are indicators for evaluating the performance of the classification model, showed a distribution of 0.7-0.9 for each class.

  • PDF

고속 객체 검출을 위한 적분 히스토그램 기반 프레임워크 (Integral Histogram-based Framework for Rapid Object Tracking)

  • 고재필;안정호;홍원기
    • 한국산업정보학회논문지
    • /
    • 제20권2호
    • /
    • pp.45-56
    • /
    • 2015
  • 본 논문에서는 스마트폰 카메라의 객체기반 자동초점 기능을 위해, 움직이는 물체의 고속 추적 방법을 제안한다. 사양이 낮은 플랫폼에서의 비-학습 제약을 고려하여 히스토그램 특징 기반의 슬라이딩 윈도우 검출 기법을 사용한다. 각 부분 윈도우에 대한 히스토그램의 계산 시간문제는 적분 히스토그램을 통해 해결한다. 본 논문에서는 지역적 후보 검출, 적응적 템플릿 크기 방법을 제안한다. 또한 추적 위치의 안정화를 위해 정합 함수에 안정화 항을 추가하는 기법을 제안한다. 자체 수집한 데이터에 대한 실험결과는 PC 환경에서 초당 100 프레임 수준의 높은 처리 속도 달성을 보여주었다.

신세대 학습자 특성 분석 : 대학생의 사이버 커뮤니케이션 이용 특성 (Analysis of New Generation Learner Characteristics : College Students Use of Cyber Communication)

  • 이옥화
    • 컴퓨터교육학회논문지
    • /
    • 제8권4호
    • /
    • pp.25-36
    • /
    • 2005
  • 2004년도 11월에 전국의 850명의 대학생의 사이버 커뮤니케이션의 이용 특성을 분석하여 이러닝 환경 구축에 관한 제언을 하였다. 학생들은 사이버 커뮤니케이션 방법 중 휴대폰 문자보내기, 인터넷 메신저 사용하기, 블로그나 미니홈피 사용하기, 이메일 보내기의 순으로 사용을 선호하였다. 57.5%의 대학생이 이메일을 사용하지 않았으며, 특히 남학생(61.9%)과 어린 학생일수록(69.4%) 그런 경향이 두드러졌다. 블로그를 소유한 사람의 비율은 74%인데, 연령이 어릴수록 많았다. 이메일의 사용은 줄고, 이동성, 동시성 접속성과 사용의 용이성이 있는 저렴한 가격의 매체의 사용이 늘고 있다. 이러닝을 위한 제언은 1) 이메일의 대체 매체를 적극 활용, 2) 접속성의 제공, 3) 동시성/즉각성의 제공, 4) 휴대폰, 블로그, 미니홈피와 같은 매체를 적극적으로 교육에 활용, 5) 텍스트정보의 멀티미디어화이다.

  • PDF

하계의 일 최고 오존농도 예측을 위한 신경망모델의 개발 (Development of Neural Network Model for Pridiction of Daily Maximum Ozone Concentration in Summer)

  • 김용국;이종범
    • 한국대기환경학회지
    • /
    • 제10권4호
    • /
    • pp.224-232
    • /
    • 1994
  • A new neural network model has been developed to predict short-term air pollution concentration. In addition, a multiple regression model widely used in statistical analysis was tested. These models were applied for prediction of daily maximum ozone concentration in Seoul during the summer season of 1991. The time periods between May and September 1989 and 1990 were utilized to train set of learning patterns in neural network model, and to estimate multiple regression model. To evaluate the results of the different models, several Performance indices were used. The results indicated that the multiple regression model tended to underpredict the daily maximum ozone concentration with small r$^{2}$(0.38). Also, large errors were found in this model; 21.1 ppb for RMSE, 0.324 for NMSE, and -0.164 for MRE. On the other hand, the results obtained from the neural network model were very promising. Thus, we can know that this model has a prominent efficiency in the adaptive control for the non-linear multi- variable systems such as photochemical oxidants. Also, when the recent new information was added in the neural network model, prediction accuracy was increased. From the new model, the values of RMSE, NMSE and r$^{2}$ were 13.2ppb, 0.089, 0.003 and 0.55 respectively.

  • PDF

웹 기반의 시각적 스트레스 증후군 선별에 관한 연구 (A Study of Web based Screening for Visual Stress Syndrome)

  • 장영건
    • 대한인간공학회지
    • /
    • 제28권4호
    • /
    • pp.91-99
    • /
    • 2009
  • A visual stress may provoke migraine, cephalalgia, intervene a long term reading and learning capability and reduce the productivity of a laborer who uses computer for a good while. In this study, a web based screening tool for visual stress was developed, it applied to 72 visually normal young persons in parallel with traditional questionnaire about symptoms of visual stress. To estimate visual stress, It is proposed to measure difference of visual search time of Han Gul characters as targets between visual stressful pattern and non-visual stressful pattern as global background of characters. As a result of test, 5 subjects were screened whose mean visual search time in the stressful pattern with 6 CPD spatial frequency increase significantly more than in none stressful pattern(t-test, one-tailed, p=$1.0407\times10^{-11}$). 2 of them were diagnosed as mild visual stress syndrome in the clinic. None had visual distortion and 50% of subjects had visual discomfort in results of the questionnaire. Correlation between visual search time increase rate and normalized number of visual discomfort may be slight(C. coefficient=0.1355). This study may contribute as a basic research for screening person with migraine, visual reading disabilities and VDT syndrome. It is required to apply the test to visual reading disabilities, migraine with visual distortion and/or visual discomfort as well as normal person for higher reliability of the screening tool.