• International Journal of Computer Science & Network Security
• /
• 제21권6호
• /
• pp.77-88
• /
• 2021

Heart Failure represents a critical pathological case that is challenging to predict and discover at an early age, with a notable increase in morbidity and mortality. Machine Learning and Neural Network techniques play a crucial role in predicting heart attacks, diseases and more. These techniques give valuable perspectives for clinicians who may then adjust their diagnosis for each individual patient. This paper evaluated neural network models for heart attacks predictions. Several online learning methods were investigated to automatically and accurately predict heart attacks. The UCI dataset was used in this work to train and evaluate First Order and Second Order Online Learning methods; namely Backpropagation, Delta bar Delta, Levenberg Marquardt and QuickProp learning methods. An optimizer technique was also used to minimize the random noise in the database. A regularization concept was employed to further improve the generalization of the model. Results show that a three layers' NN model with a Backpropagation algorithm and Nadam optimizer achieved a promising accuracy for the heart attach prediction tasks.

• International Journal of Computer Science & Network Security
• /
• 제21권6호
• /
• pp.200-206
• /
• 2021

Fraud in e-commerce transaction increased in the last decade especially with the increasing number of online stores and the lockdown that forced more people to pay for services and groceries online using their credit card. Several machine learning methods were proposed to detect fraudulent transaction. Neural networks showed promising results, but it has some few drawbacks that can be overcome using optimization methods. There are two categories of learning optimization methods, first-order methods which utilizes gradient information to construct the next training iteration whereas, and second-order methods which derivatives use Hessian to calculate the iteration based on the optimization trajectory. There also some training refinements procedures that aims to potentially enhance the original accuracy while possibly reduce the model size. This paper investigate the performance of several NN models in detecting fraud in e-commerce transaction. The backpropagation model which is classified as first learning algorithm achieved the best accuracy 96% among all the models.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제17권10호
• /
• pp.2809-2821
• /
• 2023

Effective recommendation of similar business groups is a critical factor in obtaining market information for companies. In this study, we propose a novel method for enhancing similar business group recommendation by incorporating derivative criteria and web crawling. We use employment announcements, employment incentives, and corporate vocational training information to derive additional criteria for similar business group selection. Web crawling is employed to collect data related to the derived criteria from 'credit jobs' and 'worknet' sites. We compare the efficiency of different datasets and machine learning methods, including XGBoost, LGBM, Adaboost, Linear Regression, K-NN, and SVM. The proposed model extracts derivatives that reflect the financial and scale characteristics of the company, which are then incorporated into a new set of recommendation criteria. Similar business groups are selected using a Euclidean distance-based model. Our experimental results show that the proposed method improves the accuracy of similar business group recommendation. Overall, this study demonstrates the potential of incorporating derivative criteria and web crawling to enhance similar business group recommendation and obtain market information more efficiently.

• 대한의용생체공학회:의공학회지
• /
• 제31권1호
• /
• pp.40-49
• /
• 2010

A rule weight -based fuzzy classification model is proposed to analyze the patterns of admission-discharge of patients as a previous research for differential diagnosis of dyspnea. The proposed model is automatically generated from a labeled data set, supervised learning strategy, using three procedure methodology: i) select fuzzy partition regions from spatial distribution of data; ii) generate fuzzy membership functions from the selected partition regions; and iii) extract a set of candidate rules and resolve a conflict problem among the candidate rules. The effectiveness of the proposed fuzzy classification model was demonstrated by comparing the experimental results for the dyspnea patients' data set with 11 features selected from 55 features by clinicians with those obtained using the conventional classification methods, such as standard fuzzy classifier without rule weights, C4.5, QDA, kNN, and SVMs.

• International Journal of Aeronautical and Space Sciences
• /
• 제9권1호
• /
• pp.100-110
• /
• 2008

It is not easy to monitor and identify all engine faults and conditions using conventional fault detection approaches like the GPA (Gas Path Analysis) method due to the nature and complexity of the faults. This study therefore focuses on a model based diagnostic method using Neural Network algorithms proposed for fault detection on a turbo shaft engine (PW 206C) selected as the power plant for a tilt rotor type unmanned aerial vehicle (Smart UAV). The model based diagnosis should be performed by a precise performance model. However component maps for the performance model were not provided by the engine manufacturer. Therefore they were generated by a new component map generation method, namely hybrid method using system identification and genetic algorithms that identifies inversely component characteristics from limited performance deck data provided by the engine manufacturer. Performance simulations at different operating conditions were performed on the PW206C turbo shaft engine using SIMULINK. In order to train the proposed BPNN (Back Propagation Neural Network), performance data sets obtained from performance analysis results using various implanted component degradations were used. The trained NN system could reasonably detect the faulted components including the fault pattern and quantity of the study engine at various operating conditions.

• Composites Research
• /
• 제35권5호
• /
• pp.317-321
• /
• 2022

본 연구에서는 머신 러닝을 통해 하중 유형에 따른 구간을 나누어 각 하중 유형에 강한 적층 각도 순서가 배치되는 PIC 설계 방법이 범퍼 빔에 적용되었다. 머신 러닝을 적용하기 위한 학습 데이터의 입력 값과 라벨은 각각 전체 요소 중 일부인 참조 요소의 좌표와 하중 유형으로 정의되었다. 좌표 값을 나타내는 방법인 2D 표현 방법과 3D 표현 방법을 비교하기 위하여 각각의 방법으로 학습 데이터 생성 및 머신 러닝 모델이 학습되었다. 2D 표현 방법은 유한요소 모델을 각 면으로 나누고 그에 따른 학습 데이터 생성 및 머신 러닝 모델을 학습시키는 방법이며, 3D 표현 방법은 유한요소 모델 전체에서 학습 데이터를 생성하여 하나의 머신 러닝 모델을 학습시키는 방법이다. 머신 러닝 모델의 성능에 영향을 미치는 하이퍼파라미터는 베이지안 알고리즘을 통해 최적 값으로 튜닝되었으며, 튜닝 된 모델 중 k-NN 분류 방법이 가장 높은 예측률과 AUC-ROC로 나타났다. 그리고 2D 표현 방법과 3D 표현 방법 중 3D 표현 방법이 더 높은 성능을 보였다. 튜닝 된 머신 러닝 모델을 통해 예측된 하중 유형 데이터가 유한요소 모델에 매핑되었으며, 유한요소 해석을 통해 비교 검증되었다. 3D 표현 방법의 머신 러닝 모델로 설계된 PIC 방법이 강도 측면에서 더 우수함이 검증되었다.

• 제어로봇시스템학회논문지
• /
• 제11권3호
• /
• pp.207-213
• /
• 2005

This paper deals with an autonomous flight guidance and control algorithm design for TR301 tilt-rotor airplane under development by Korea Aerospace Research Institute for simulation purpose. The objective of this study is to design autonomous flight algorithm in which the tilt-rotor airplane should follow the given waypoints precisely. The approach to this objective in this study is that, first of all, model-based inversion is applied to the highly nonlinear tilt-rotor dynamics, where the tilt-rotor airplane is assumed to fly at helicopter flight mode(nacelle angle=0 deg), and then the control algorithm, based on classical control, is designed to satisfy overall system stabilization and precise waypoint following performance. Especially, model uncertainties due to the tiltrotor model itself and inversion process are adaptively compensated in a simple neural network(Sigma-Phi NN) for performance robustness. The designed algorithm is evaluated in the tilt-rotor nonlinear airplane in helicopter flight mode to analyze the following performance for given waypoints. The simulation results show that the waypoint following responses for this algorithm are satisfactory, and control input responses are within control limits without saturation.

• 로봇학회논문지
• /
• 제19권2호
• /
• pp.203-212
• /
• 2024

Simultaneous motion is essential in the activities of daily living (ADL). For motion intention recognition, surface electromyogram (sEMG) and corresponding motion label is necessary. However, this process is time-consuming and it may increase the burden of the user. Therefore, we propose a simultaneous motion recognition framework using data augmentation based on muscle activation model. The model consists of multiple point sources to be optimized while the number of point sources and their initial parameters are automatically determined. From the experimental results, it is shown that the framework has generated the data which are similar to the real one. This aspect is quantified with the following two metrics: structural similarity index measure (SSIM) and mean squared error (MSE). Furthermore, with k-nearest neighbor (k-NN) or support vector machine (SVM), the classification accuracy is also enhanced with the proposed framework. From these results, it can be concluded that the generalization property of the training data is enhanced and the classification accuracy is increased accordingly. We expect that this framework reduces the burden of the user from the excessive and time-consuming data acquisition.

• 전자공학회논문지CI
• /
• 제44권1호
• /
• pp.56-63
• /
• 2007

본 논문은 신경망 (neural network: NN)과 mean-shift알고리즘을 이용하여 복잡한 배경에서 사용자의 눈을 정확히 추출하고 추적할 수 있는 눈 추적 시스템을 제안한다. 머리의 움직임에 강건한 시스템을 개발하기 위해서 먼저 피부색 모델과 연결 성분분석을 이용하여 얼굴영역을 추출한다. 그 다음 신경망기반의 텍스처 분류기를 이용하여 얼굴 영역(face region)을 눈 영역(eye region)과 비눈 영역(non-eye region)으로 구분함으로써 눈을 찾는다. 이러한 눈 검출 방법은 안경의 착용 유무에 상관없이 사용자의 눈 영역을 정확히 검출 할 수 있게 한다. 일단 눈 영역이 찾아지면 이후 프레임에서의 눈 영역은 mean-shift알고리즘에 의해 정확하게 추적된다. 제안된 시스템의 효율성을 검증하기 위해서 제안된 시스템은 눈의 움직임을 이용한 인터페이스 시스템에 적용되었고, 이 인터페이스를 이용한 'aliens game'이 구현되었다. 25명의 사용자에 대해 실험한 결과는 제안된 시스템이 보다 편리하고 친숙한 인터페이스로 활용될 수 있다는 것을 보여주었으며, 또한 $320{\times}240$ 크기의 영상을 초당 30프레임의 빠른 속도로 처리함으로써 실시간 시스템에 적용될 수 있음을 보여주었다.

• 한국산학기술학회논문지
• /
• 제13권3호
• /
• pp.1255-1262
• /
• 2012

현재 가동 중인 월성 원자력 발전의 핵연료 교체로봇 시스템을 살펴보면 핵연료 교환에 필요한 구동압력 제어를 위해 PI제어기를 사용한다. PI제어는 구조가 간단하고 이득 설정을 통해 시스템 요구조건에 만족하는 제어 성능을 낼 수 있지만 밸브와 관로 등의 파라미터 변화로부터 적절한 이득 변경 없이 안정한 제어가 힘들다. 이러한 문제를 해결하기 위해 PI제어기 이득을 동적으로 변경 하거나 PI제어기 출력을 보상하도록 제어기를 구성하는 것이 바람직하다. 본 연구개발의 목적은 파라미터 변화에도 안정한 제어가 가능하도록 제어기를 설계하여 오차와 진동현상을 줄이는데 있다. 제안한 PI/NN제어 기법은 PI제어기와 신경회로망 제어기를 병렬 결합한 구조로 신경회로망 제어기가 PI제어기 출력을 보상하여 파라미터 변화에 강인하도록 설계 하였다. 제어기의 성능평가를 위해 직접 실 공정에 테스트하기가 힘들기 때문에 공정의 특성을 반영하여 모델링한 시뮬레이터를 개발하였고, 시뮬레이션 결과를 실 공정데이터와 비교하여 공정 특성을 모사함을 보였으며, 파라미터 변화에 PI/NN제어기가 오차 및 진동현상을 줄이는 것을 확인 하였다. 또한, 실 공정에서 사용 중인 PI제어기를 주 제어기로 사용하면서 파라미터 변화에 대한 비선형성을 보상하는 제어기 역할을 하기 때문에 신경회로망을 단독으로 사용하였을 때 보다 더 신뢰성 있고 안정적인 제어가 가능하다.