데이터와 인공신경망 능력 계산

이덕균,박지은 한국정보통신학회 2022 한국정보통신학회논문지 Vol.26 No.1

Recently, various uses of artificial intelligence have been made possible through the deep artificial neural network structure of machine learning, demonstrating human-like capabilities. Unfortunately, the deep structure of the artificial neural network has not yet been accurately interpreted. This part is acting as anxiety and rejection of artificial intelligence. Among these problems, we solve the capability part of artificial neural networks. Calculate the size of the artificial neural network structure and calculate the size of data that the artificial neural network can process. The calculation method uses the group method used in mathematics to calculate the size of data and artificial neural networks using an order that can know the structure and size of the group. Through this, it is possible to know the capabilities of artificial neural networks, and to relieve anxiety about artificial intelligence. The size of the data and the deep artificial neural network are calculated and verified through numerical experiments. 최근 인공지능의 다양한 활용은 기계학습의 딥 인공신경망 구조를 통해 가능해졌으며 인간과 같은 능력을 보여주고 있다. 불행하게도 딥 구조의 인공신경망은 아직 정확한 해석이 이루어지고 있지 못하고 있다. 이러한 부분은 인공지능에 대한 불안감과 거부감으로 작용하고 있다. 우리는 이러한 문제 중에서 인공신경망의 능력 부분을 해결한다. 인공신경망 구조의 크기를 계산하고, 그 인공신경망이 처리할 수 있는 데이터의 크기를 계산해 본다. 계산의 방법은 수학에서 쓰이는 군의 방법을 사용하여 데이터와 인공신경망의 크기를 군의 구조와 크기를 알 수 있는 Order를 이용하여 계산한다. 이를 통하여 인공신경망의 능력을 알 수 있으며, 인공지능에 대한 불안감을 해소할 수 있다. 수치적 실험을 통하여 데이터의 크기와 딥 인공신경망을 계산하고 이를 검증한다.

딥러닝 신경망을 이용한 신용카드 부도위험 예측의 효용성 분석

윤종문 한국금융학회 2019 금융연구 Vol.33 No.1

This study aims to discuss the usefulness of the deep learning neural network and the possibility of the deep learning neural network analysis in judging credit information by using credit card default data. Deep learning neural network analysis in the financial sector excluding the current stock price prediction model is under limited research. It is mainly used for upgrading models of the credit rating (Kvamme et al., 2016, 2018; Tran, 2016; Luo, 2017) and the delinquency rate (Sirignano et al., 2018). In the credit card market, it is focused on credit card issuance and fraud detection model (Ramanathan, 2014, Niimi, 2015). As mentioned earlier, there has not been much analysis of deep learning neural network using financial market data. This is because the study of deep learning neural networks is actively carried out mainly in the field of computer science such as image, speech recognition, natural language processing. Additionally, Researchers in the financial sector have difficulty learning deep learning algorithms and setting up a computer runtime environment. It is also difficult to apply the algorithm to financial data due to lower dimension than the image. Nowadays, financial companies have been interested in machine learning and are increasing their recruitment, but it is still in the stage of verifying the possibility of deep learning neural network. Therefore, This study examines the possibility of improving the accuracy of credit card default risk prediction by using a deep learning neural network algorithm. To do this, we use existing machine learning algorithms (Logistic, SVM, Random Forest, Lasso, etc.) as a comparison index for performance check of deep learning neural network analysis. Firstly, the deep learning neural network is constructed with two hidden layers and five neurons, and derives the prediction accuracy according to the activation function and the initial value setting method. There are Sigmoid, ReLU, tanh and Maxout as active functions, and random value, Xavier, RBM, He’s as initialization methods. Based on this, we compare the accuracy of existing machine learning algorithms. As a result, the deep learning neural network analysis showed performance improvement between 0.6% and 6.6%p compared to the existing machine learning algorithms (Logistic, SVM, Random Forest, Lasso, etc.). Among these results, the active function and the initial value setting method with the highest prediction accuracy are ReLU (rectified linear units) and Xavier initialization. However, there is no significant improvement in performance with increasing number of hidden layers and neurons up to 10 and 25, respectively. Also, the dropout and CNN (convolution neural network) models, which showed high performance in the field of image identification, showed no significant difference in prediction accuracy. Nevertheless, it could be interpreted that the increase of hidden layers can improve the accuracy of estimation because the highest accuracy (0.8161) and the AUC (0.7726) are observed for 10 hidden nodes and 15 neurons. However, we can’t say that accuracy increases linearly by the number of hidden layers and neurons. These limitation could be due to the quantitative and qualitative limitations of the credit card data used here. We did not use recurrent neural network (RNN) and long-short term memory (LSTM) models since the personal default data for credit card used in this study is cross-sectional data. These method are for Time-Series data. Therefore, it is expected that it will be able to obtain better results in identification problems (credit rating, delinquency rate, interest rate calculation) of present various financial markets if these deep learning neural network methodologies are applied through big data including time series data. This study can be turned into a question of how deep learning analysis can lower the default risk and delinquency rate by using financial data from a practical point of ... 본 연구는 국내․외 금융시장에서 아직 활성화되지 못한 딥러닝 신경망(deep learning neural network) 알고리즘을 이용해 신용카드 부도위험 예측의 정확도 향상 가능성에 대해서 점검한다. 이를 위해 기존 머신러닝 알고리즘(Logistic, SVM, Random Forest, Lasso 등)을 딥러닝 신경망 분석의 성능 점검을 위한 비교 지표로 활용한다. 우선, 딥러닝 신경망은 두 개의 은닉층(hidden layers)과 다섯 개의 뉴런(neuron)으로 구축하고, 활성함수(activation function)와 초기값(initial value) 설정방법에 따른 예측정확도를 도출한다. 그 결과 딥러닝 신경망 분석이 기존 머신러닝 알고리즘 보다 최소 0.6%p에서 최대 6.6%p 성능이 향상된 것으로 나타났다. 이 중 가장 높은 예측 정확도를 보인 활성함수와 초기값 설정방식은 ReLU(rectified linear units)와 Xavier(2010)이고 이를 기준으로 은닉층과 뉴런의 수를 각각 최대 10개와 25개까지 늘려 분석한 결과에서도 유사한 결과가 나타났다. 다만, 기존 연구에서와 같이 은닉층과 뉴런의 수의 증가에 따른 뚜렷한 성능의 향상은 나타나지 않았다. 또한, 이미지 식별 분야에서 높은 성능을 보였던 Dropout과 CNN(convolution neural network) 모델도 예측 정확도에서 큰 차이를 보이지 않았다. 이는 여기에서 사용된 신용카드 데이터가 다수 픽셀(pixel)로 이루어진 이미지 데이터와 비교해 양적․질적 한계가 있기 때문으로 판단된다. 한편, 본 연구에서 사용된 개인의 신용카드 부도 데이터는 횡단면 자료이기 때문에 시계열 데이터에서 높은 성능을 나타내는 RNN(recurrent neural network) 및 LSTM(Long- Short Term Memory) 등의 딥러닝 신경망 알고리즘을 사용하지는 않았다. 따라서 추후 시계열 자료가 포함된 빅데이터를 통해 이들 딥러닝 신경망 방법론을 적용한다면, 현재의 다양한 금융시장의 식별문제(신용등급, 연체율, 금리산정)에 있어 보다 향상된 결과를 도출할 수 있을 것으로 기대된다.

Geometry Characteristics Prediction of Single Track Cladding Deposited by High Power Diode Laser Based on Genetic Algorithm and Neural Network

Huaming Liu,Xunpeng Qin,Song Huang,Lei Jin,Yongliang Wang,Kaiyun Lei 한국정밀공학회 2018 International Journal of Precision Engineering and Vol.19 No.7

This paper aims to establish a correlation between the process parameters and geometrical characteristics of the sectional profile of the single track cladding deposited by high power diode laser with rectangle beam spot. By applying the genetic algorithm and back propagation neural network, a nonlinear model for predicting the geometry features of the single track cladding is developed. A full factorial design method is used to conduct the experiments, and the experimental results are chosen randomly as training dataset and testing dataset for the neural network. Three main input variables such as laser power, scanning speed, and powder thickness were considered. The performance of the genetic algorithm and back propagation artificial neural network was compared to that of the standard back propagation neural network. To improve the accuracy of the neural network, one-hidden-layer and double-hidden-layer neural network with different architectures were performed. Further, one-output and multi-output neural network are also trained and tested. The results indicate that, by using genetic algorithm, the prediction accuracy of the neural network is significantly improved. Meanwhile, the double-hidden-neural network has higher prediction accuracy than the one-hidden-layer-neural network, while the one-output-neural network has higher prediction accuracy than the multi-output-neural network.

인공신경망을 이용한 KOMPSAT-3/3A/5 영상으로부터 자연림과 인공림의 분류

이용석,박숭환,정형섭,백원경 대한원격탐사학회 2018 大韓遠隔探査學會誌 Vol.34 No.6

Natural forests are un-manned forests where the artificial forces of people are not applied to the formation of forests. On the other hand, artificial forests are managed by people for their own purposes such as producing wood, preventing natural disasters, and protecting wind. The artificial forests enable us to enhance economical benefits of producing more wood per unit area because it is well-maintained with the purpose of the production of wood. The distinction surveys have been performed due to different management methods according to forests. The distinction survey between natural forests and artificial forests is traditionally performed via airborne remote sensing or in-situ surveys. In this study, we suggest a classification method of forest types using satellite imagery to reduce the time and cost of in-situ surveying. A classification map of natural forest and artificial forest were generated using KOMPSAT-3, 3A, 5 data by employing artificial neural network (ANN). And in order to validate the accuracy of classification, we utilized reference data from 1/5,000 stock map. As a result of the study on the classification of natural forest and plantation forest using artificial neural network, the overall accuracy of classification of learning result is 77.03% when compared with 1/5,000 stock map. It was confirmed that the acquisition time of the image and other factors such as needleleaf trees and broadleaf trees affect the distinction between artificial and natural forests using artificial neural networks. 자연림은 산림의 조성 과 보육 등에 인공적인 사람의 힘이 가해지지 않은 자연 상태의 산림이다. 반면 인공림은 사람이 조성 및 보육관리 하는 숲으로 목재생산, 자연재해 예방, 방풍 등의 목적을 가지는 산림이다. 인공림은 목재생산 등 인간이 목적을 가지고 관리하여 단위 면적당 더 많은 목재를 생산할 수 있는 경제적 장점도 가지고 있다. 자연림과 인공림의 구분은 산림 형태의 관리 방법과 목정이 상이하여 산림조사에서 기본적으로 조사하는 요소이며, 자연림과 인공림의 구분은 항공사진 판독과 현지조사 등의 절차를 통해 이루어진다. 본 연구에서는 자연림과 인공림의 분류에 KOMPSAT-3, 3A, 5 위성 영상데이터에 인공신경망(Artificial Neural Network: ANN)을 적용하여 자연림과 인공림의 분류도를 만들고, 산림청의 1/5,000임상도의 자연림과 인공림 분류도와 비교하여 평가하였다. 인공신경망을 이용한 산림의 자연림과 인공림 구분의 연구를 진행한 결과, 1/5,000 임상도와 비교했을 때, 학습결과 분류 전체 정확도는 77.03%이다. 영상의 획득 시기와 산림의 침엽수와 활엽수 등 기타요인이 인공신경망을 이용한 산림의 인공림과 자연림의 구분에 많은 영향을 미치는 것을 확인하였다.

배경기계학습을 사용한 침입탐지시스템

김형훈(Hyung-Hoon Kim),조정란(Jeong-Ran Cho) 한국컴퓨터정보학회 2019 韓國컴퓨터情報學會論文誌 Vol.24 No.5

The existing subtract image based intrusion detection system for CCTV digital images has a problem that it can not distinguish intruders from moving backgrounds that exist in the natural environment. In this paper, we tried to solve the problems of existing system by designing real - time intrusion detection system for CCTV digital image by combining subtract image based intrusion detection method and background learning artificial neural network technology. Our proposed system consists of three steps: subtract image based intrusion detection, background artificial neural network learning stage, and background artificial neural network evaluation stage. The final intrusion detection result is a combination of result of the subtract image based intrusion detection and the final intrusion detection result of the background artificial neural network. The step of subtract image based intrusion detection is a step of determining the occurrence of intrusion by obtaining a difference image between the background cumulative average image and the current frame image. In the background artificial neural network learning, the background is learned in a situation in which no intrusion occurs, and it is learned by dividing into a detection window unit set by the user. In the background artificial neural network evaluation, the learned background artificial neural network is used to produce background recognition or intrusion detection in the detection window unit. The proposed background learning intrusion detection system is able to detect intrusion more precisely than existing subtract image based intrusion detection system and adaptively execute machine learning on the background so that it can be operated as highly practical intrusion detection system.

적응학습 퍼지-신경회로망에 의한 IPMSM의 최대토크 제어

정동화,고재섭,최정식,Chung, Dong-Hwa,Ko, Jae-Sub,Choi, Jung-Sik 한국조명전기설비학회 2007 조명·전기설비학회논문지 Vol.21 No.5

IPMSM은 하중에 비하여 고출력으로 인하여 전기자동차에 널리 보급되고 있다. 본 논문은 적응 학습 퍼지-신경회로망과 ANN을 이용한 IPMSM드라이브의 최대토크 제어를 제시한다. 이러한 제어 방법은 인버터의 정격전류 및 전압값의 범위를 고려한 전속도 영역에 적용 된다. 본 논문은 적응학습 퍼지-신경회로망을 이용하여 IPMSM의 속도제어와 ANN을 이용하여 속도를 추정을 제시한다. 신경회로망의 역전파 알고리즘은 전동기 속도의 실시간 추정을 제시하는데 사용된다. 제시된 제어 알고리즘은 적응학습 퍼지-신경회로망과 ANN 제어기를 IPMSM 드라이브에 적용된다. 최대토크에 의해 제어된 동작 특성은 세부적으로 실험한다. 또한 본 논문은 적응 학습 퍼지 신경회로망과 ANN의 효과를 결과 분석을 통해 제시한다. Interior permanent magnet synchronous motor(IPMSM) has become a popular choice in electric vehicle applications, due to their excellent power to weight ratio. This paper proposes maximum torque control of IPMSM drive using adaptive learning fuzzy neural network and artificial neural network. This control method is applicable over the entire speed range which considered the limits of the inverter's current and voltage rated value. This paper proposes speed control of IPMSM using adaptive learning fuzzy neural network and estimation of speed using artificial neural network. The back propagation neural network technique is used to provide a real time adaptive estimation of the motor speed. The proposed control algorithm is applied to IPMSM drive system controlled adaptive learning fuzzy neural network and artificial neural network, the operating characteristics controlled by maximum torque control are examined in detail. Also, this paper proposes the analysis results to verify the effectiveness of the adaptive learning fuzzy neural network and artificial neural network.

인공지능 이용 범죄예측 기법과 불심검문 등에의 적용에 관한 고찰

양종모(Yang Jong Mo) 대검찰청 2016 형사법의 신동향 Vol.0 No.51

빅 데이터 시대의 본격적 개막에 따라 빅 데이터를 각 분야에 응용하려는 다양한 시도가 진행되고 있다. 이러한 빅 데이터의 응용에 있어서 핵심적 요소는 인공지능을 이용한 분석기법인데, 범죄예측 분야도 예외는 아니어서 빅 데이터와 인공지능을 이용한 범죄예측의 실용적 모델들이 속속 등장하고 있다. 이러한 인공지능과 빅 데이터를 이용한 범죄예측 기법은 1980년대의 인공지능 전성시대의 기법과는 다른 머신러닝이라는 새로운 기법 때문에 가능하게 된 것인데, 사전에 범죄발생 징후가 높은 지역과 시간을 미리 예측하여 선정하고, 순찰활동 등 경찰력을 그 지점에 집중토록 함으로써 경찰의 예방활동에 현실적이고도, 지대한 공헌을 하고 있다. 이런 시스템을 범죄예방 활동에 도입한 미국의 여러 도시에서 괄목한 만한 성과가 나옴으로써 ‘마이너리티 리포트’라는 영화에서나 가능했던 범죄예측과 사전 진압의 스토리가 멀지 않은 장래에 실현될 가능성을 높여주고 있다. 본고에서는 이러한 인공지능을 이용한 범죄예측 기법에 대한 일반적 소개와 더불어, 그것을 실제 활용함에 있어 우리 법 현실에서 어떠한 법적, 기술적, 방법론적인 문제를 야기할 것인지, 그런 문제를 현행법 체계 하에서 어떻게 규율할 수 있는지에 대하여 불심검문에의 적용을 중심으로 검토할 예정이다. 불심검문 제도의 특수성 때문에 그 규율에 있어서 다른 분야와는 다른 특성이 나타나기도 하는데, 보다 범용적인 형태의 규율은 차후의 과제로 남기고, 논의의 범위를 인공지능 분석기법에 의한 범죄예측 결과의 불심검문 적용에 한정하기로 한다. 머신러닝 분석기법 중 인공신경망은 사용자에게 일종의 블랙박스와 같고, 불투명하기 때문에 결론 도출과정을 설명할 수 없다는 특성을 가지고 있는데, 이러한 특성 때문에 머신러닝 분석기법, 나아가 인공지능 분석기법이 적용된 예측결과를 과학적 증거방법으로 허용하기 위해서는 여러가지 법적 검토가 뒤따라야 할 것이다. 나아가 인공지능에 대한 일반적 허용과 관련하여서도 그와 관련된 법적·제도적 측면의 논의 외에 법철학적 차원 등 다양한 형태의 고찰이 필요하다는 점은 분명하다. After historic man-versus-machine match in Seoul, people became aware of rapid advance of artificial intelligence. Artificial intelligence is based on the assumption that the process of human thought can be mechanized. The Beginning of artificial intelligence is set in 1956. Watson, the artificial intelligence program created by I. B. M. is able to understand natural language queries and answer question. Machine Learning is the hot new phase of artificial intelligence. This has become practical only recently with the development of big data. Applying machine learning is a big switch from traditional police dispatching. The use of artificial intelligence to predict crime has only recently emerged. For example Predpol software is a powerful tool that allows law enforcement to shift from reactive to preventive policing. It’s algorithm uses advanced mathematics and machine learning to generate predictions using type, place, and time of past crimes. It is based on the idea that analyzing large amount of crime data, you can more accurately predict where and when crimes may happen. This article reviews the admissibility of predictive policing based on artificial intelligence. The standard rule on admissibility of new scientific evidence is Frye and Daubert test. But the reliability of machine learning should be differently assessed. The heart of machine learning is artificial neural network. Artificial neural network can learn, but acts as a black box. So artificial neural network can’t explain how it arrives at a particular solution. This opaqueness affects the admissibility of crime prediction based on machine learning. If we cannot see how the network derives its results it produces, we cannot accept the results. The validation of results is very important factor in admissibility of scientific evidence in criminal trials. But predictive policing based on artificial intelligence will impact reasonable suspicion analysis of police stop. Despite that predictive policing based on artificial intelligence lacks reliability and transparency, it provides legitimacy in police stop.

인공 면역망과 신경회로망을 이용한 자율이동로봇 주행

이동제,김인식,이민중,최영규 대한전기학회 2003 전기학회논문지 D Vol.53 No.8

The acts of biological immune system are similar to the navigation for autonomous mobile robots under dynamically changing environments. In recent years, many researchers have studied navigation algorithms using artificial immune networks. Conventional artificial immune algorithms consist of an obstacle-avoidance behavior and a goal-reaching behavior. To select a proper action, the navigation algorithm should combine the obstacle-avoidance behavior with the goal-reaching behavior. In this paper, the neural network is employed to combine the behaviors. The neural network is trained with the surrounding information. the outputs of the neural network are proper combinational weights of the behaviors in real-time. Also, a velocity control algorithm is constructed with the artificial immune network. Through a simulation study and experimental results for a autonomous mobile robot, we have shown the validity of the proposed navigation algorithm.

An Artificial Neural Network Learning Fuzzy Membership Functions for Extracting Color Sketch Features

조성목(Sung-Mok Cho),조옥래(Ok-Lae Cho) 한국컴퓨터정보학회 2006 韓國컴퓨터情報學會論文誌 Vol.11 No.3

본 논문에서는 칼라 영상의 스케치 특징점을 추출하기 위해 퍼지신경회로망을 이용하는 방법에 대하여 설명한다. 이 신경회로망은 스케치 특징점 추출을 위한 퍼지 소속함수를 학습시킴으로써 적절한 국부 임계치를 획득할 수 있도록 구성된다. 제안한 퍼지신경회로망의 입출력 소속함수는 표준영상으로부터 추출된 최적의 특징점 추출결과를 기반으로 구성하여 학습 데이타로 사용된다. 학습에 사용된 퍼지입력변수는 디지털 영상에서의 특징점 추출 시 국부영역 밝기를 잘 반영할 뿐만 아니라 특징점 추출성능이 매우 우수한 특성이 있으며, 이들 입력변수의 소속함수를 신경회로망으로 학습시킴으로써 매우 효과적이고 신속하게 스케치 특징점들을 추출할 수 있다. 실험결과, 소속함수로 학습된 신경회로망으로부터 얻어진 임계치를 사용한 특징점 추출이 다양한 영상에 대하여 매우 우수함을 보였다. This paper describes the technique which utilizes a fuzzy neural network to sketch feature extraction in digital images. We configure an artificial neural network and make it learn fuzzy membership functions to decide a local threshold applying to sketch feature extraction. To do this, we put the learning data which is membership functions generated based on optimal feature map of a few standard images into the artificial neural network. The proposed technique extracts sketch features in an image very effectively and rapidly because the input fuzzy variable have some desirable characteristics for feature extraction such as dependency of local intensity and excellent performance and the proposed fuzzy neural network is learned from their membership functions, We show that the fuzzy neural network has a good performance in extracting sketch features without human intervention.

Prediction of persistent hemodynamic depression after carotid angioplasty and stenting using artificial neural network model

Jeon, Jin Pyeong,Kim, Chulho,Oh, Byoung-Doo,Kim, Sun Jeong,Kim, Yu-Seop Van Gorcum 2018 Clinical neurology and neurosurgery Vol.164 No.-

<P><B>Abstract</B></P> <P><B>Objectives</B></P> <P>To assess and compare predictive factors for persistent hemodynamic depression (PHD) after carotid artery angioplasty and stenting (CAS) using artificial neural network (ANN) and multiple logistic regression (MLR) or support vector machines (SVM) models.</P> <P><B>Patients and methods</B></P> <P>A retrospective data set of patients (n=76) who underwent CAS from 2007 to 2014 was used as input (training cohort) to a back-propagation ANN using TensorFlow platform. PHD was defined when systolic blood pressure was less than 90mmHg or heart rate was less 50 beats/min that lasted for more than one hour. The resulting ANN was prospectively tested in 33 patients (test cohort) and compared with MLR or SVM models according to accuracy and receiver operating characteristics (ROC) curve analysis.</P> <P><B>Results</B></P> <P>No significant difference in baseline characteristics between the training cohort and the test cohort was observed. PHD was observed in 21 (27.6%) patients in the training cohort and 10 (30.3%) patients in the test cohort. In the training cohort, the accuracy of ANN for the prediction of PHD was 98.7% and the area under the ROC curve (AUROC) was 0.961. In the test cohort, the number of correctly classified instances was 32 (97.0%) using the ANN model. In contrast, the accuracy rate of MLR or SVM model was both 75.8%. ANN (AUROC: 0.950; 95% CI [confidence interval]: 0.813–0.996) showed superior predictive performance compared to MLR model (AUROC: 0.796; 95% CI: 0.620–0.915, p<0.001) or SVM model (AUROC: 0.885; 95% CI: 0.725-0.969, p<0.001).</P> <P><B>Conclusions</B></P> <P>The ANN model seems to have more powerful prediction capabilities than MLR or SVM model for persistent hemodynamic depression after CAS. External validation with a large cohort is needed to confirm our results.</P> <P><B>Highlights</B></P> <P> <UL> <LI> Prediction of hemodynamic depression after CAS is possible using artificial neural network. </LI> <LI> Accuracy of artificial neural network was 98.7% with 0.961 of AUROC. </LI> <LI> It is feasible to use artificial neural network to predict high-risk patient of hemodynamic depression. </LI> </UL> </P>