• International Journal of Internet, Broadcasting and Communication
• /
• 제11권2호
• /
• pp.11-18
• /
• 2019

One of the key objectives of control systems is to maintain a system in a specific stable state. To achieve this goal, a variety of control techniques can be used and it is often uses a feedback control method. As known this kind of control methods requires mathematical model of the system. This article presents seesaw unstable system with two propellers which are controlled without use of a mathematical model instead. The goal was to control it using training data. For system control we use a logistic regression technique which is one of machine learning method. We tested our controller on the real model created in our laboratory and the experimental results show that instability of the seesaw system can be fixed at a given angle using the decision boundary estimated from the classification method. The results show that this control method for structural equilibrium can be used with relatively more accuracy of the decision boundary.

• 국제초고층학회논문집
• /
• 제9권4호
• /
• pp.335-342
• /
• 2020

This article comes up with a new method which is based on the visual characteristic of the objects and machine learning technology to achieve semi-automated recognition of the personnel, machine & materials of the construction sites. Balancing the real-time performance and accuracy, using Faster RCNN (Faster Region-based Convolutional Neural Networks) with transfer learning method appears to be a rational choice. After fine-tuning an ImageNet pre-trained Faster RCNN and testing with it, the result shows that the precision ratio (mAP) has so far reached 67.62%, while the recall ratio (AR) has reached 56.23%. In other word, this recognizing method has achieved rational performance. Further inference with the video of the construction of Huoshenshan Hospital also indicates preliminary success.

• Journal of information and communication convergence engineering
• /
• 제20권1호
• /
• pp.31-40
• /
• 2022

Coronary heart disease (CHD) is a comorbidity of COVID-19; therefore, routine early diagnosis is crucial. A large number of examination attributes in the context of diagnosing CHD is a distinct obstacle during the pandemic when the number of health service users is significant. The development of a precise machine learning model for diagnosis with a minimum number of examination attributes can allow examinations and healthcare actions to be undertaken quickly. This study proposes a CHD diagnosis model based on feature selection, data balancing, and ensemble-based classification methods. In the feature selection stage, a hybrid SVM-GA combined with fast correlation-based filter (FCBF) is used. The proposed system achieved an accuracy of 94.60% and area under the curve (AUC) of 97.5% when tested on the z-Alizadeh Sani dataset and used only 8 of 54 inspection attributes. In terms of performance, the proposed model can be placed in the very good category.

• 정보보호학회논문지
• /
• 제21권2호
• /
• pp.101-110
• /
• 2011

최근 Application에 대한 공격을 통하여 네트워크와 데이터베이스에 대한 방어정책인 침입탐지 룰(rule)을 무력화시키고, 침해사고를 유발한다. 이러한 공격으로부터 내부 네트워크와 데이터베이스의 안전성을 확보하기 위한 통합보안에 관한 연구가 필요하다. 본 논문에서는 침입탐지 룰을 설정한 Application에 대한 공격을 차단하기 위한 통합보안 시스템 구축에 관한 연구이다. 네트워크 기반의 공격을 탐지하여 대응하고, 내부 Integration Security System을 Virtual clustering과 Load balancing 기법으로 공격을 분산시키며, Packet 모니터링과 분석을 통하여 공격 목적지 Packet에 대한 방어정책 설정, 공격 Packet 분석, 기록, 룰 업데이트를 한다. 또한 공격 유형별 방어정책을 설정하여 Virtual Machine 분할 정책을 통한 접근 트래픽 해소, 공격차단에 적용하는 Integration Security System을 제안하고 방어를 실험한다. 본 연구 결과는 외부 해커의 공격에 대한 통합보안 방어를 위한 현실적인 자료를 제공하게 될 것이다.

• ETRI Journal
• /
• 제42권5호
• /
• pp.686-699
• /
• 2020

In this paper, we propose a supervised-learning-based spatial performance prediction (SLPP) framework for next-generation heterogeneous communication networks (HCNs). Adaptive asset placement, dynamic resource allocation, and load balancing are critical network functions in an HCN to ensure seamless network management and enhance service quality. Although many existing systems use measurement data to react to network performance changes, it is highly beneficial to perform accurate performance prediction for different systems to support various network functions. Recent advancements in complex statistical algorithms and computational efficiency have made machine-learning ubiquitous for accurate data-based prediction. A robust network performance prediction framework for optimizing performance and resource utilization through a linear discriminant analysis-based prediction approach has been proposed in this paper. Comparison results with different machine-learning techniques on real-world data demonstrate that SLPP provides superior accuracy and computational efficiency for both stationary and mobile user conditions.

• 지능정보연구
• /
• 제5권1호
• /
• pp.95-101
• /
• 1999

In this paper, a comparison between different methods to combine predictions from neural networks will be given. These methods are bagging, bumping, and balancing. Those are based on the analysis of the ensemble generalization error into an ambiguity term and a term incorporating generalization performances of individual networks. Neural Networks and AI machine learning models are prone to overfitting. A strategy to prevent a neural network from overfitting, is to stop training in early stage of the learning process. The complete data set is spilt up into a training set and a validation set. Training is stopped when the error on the validation set starts increasing. The stability of the networks is highly dependent on the division in training and validation set, and also on the random initial weights and the chosen minimization procedure. This causes early stopped networks to be rather unstable: a small change in the data or different initial conditions can produce large changes in the prediction. Therefore, it is advisable to apply the same procedure several times starting from different initial weights. This technique is often referred to as training ensembles of neural networks. In this paper, we presented a comparison of three statistical methods to prevent overfitting of neural network.

• Industrial Engineering and Management Systems
• /
• 제10권3호
• /
• pp.197-202
• /
• 2011

In traditional production lines, such as assembly lines, each worker is usually assigned to a particular fixed work, and decreasing the task to master the assigned work is valuated. However, when an imbalance exists between workers' speeds, if a worker delays the overall work in the production line, the production rate of the particular line will also decrease. To avoid this problem, the "Self-Balancing Production Line" was introduced. In this type of production line, each worker is assigned work dynamically, and when specific conditions are satisfied, production remains balanced. Characteristics of these lines that can be preempted at any place have already been analyzed by some researchers. A previous paper examined the situation in which only a single worker can process one machine and cannot preempt processing, and the improved policy of an ordinary selfbalancing production line, which specifies which stations workers can process and how workers can behave. This policy achieveda high production rate with only four stations and two workers (Buzacott, 2002). In that paper, worker processing stations and the behavior of a specific worker were limited, andthe paper focused only on specific stations and workers. Therefore, it is not applicable to any worker sequence. In this paper, we focus on other ways to decrease cycle time. In this kind of line, a worker processes at his or her speed. Therefore, if a worker is assigned stations according to his or her speed, the line can decrease cycle time. To do so, we relax the assumptions of this type of line and set a new condition. Under these conditions, we compare our results to the results of previous papers.

• 정보처리학회논문지:소프트웨어 및 데이터공학
• /
• 제7권5호
• /
• pp.177-188
• /
• 2018

최근 웹, 소셜 네트워크 서비스, 모바일, 사물인터넷 등의 ICT 기술의 발전으로 인해 처리 및 분석이 필요한 그래프 데이터의 규모가 급속하게 증가하였다. 이러한 대규모 그래프 데이터는 단일 기기에서의 처리가 어렵기 때문에 여러 기기에 나누어 분산/병렬 처리하는 것이 필요하다. 기존 그래프 처리 알고리즘들은 단일 메모리 환경을 기반으로 연구되어 분산/병렬 처리환경에 적용되기 힘들다. 이에 대규모 그래프의 보다 효과적인 분산/병렬 처리를 위해 정점 중심 방식의 그래프 처리 시스템들과, 정점 중심 방식의 단점을 보완한 블록 중심 방식의 그래프 처리 시스템들이 등장하였다. 이러한 시스템들은 초기 그래프 분할 상태가 전체 처리 성능에 상당한 영향을 미친다. 한 번에 최적의 상태로 그래프를 분할하는 것은 매우 어려운 문제이므로, 그래프 처리 시간에 점진적으로 그래프 분할 상태를 개선하는 여러 로드 밸런싱 기법들이 연구되었다. 그러나 기존 기법들은 대부분 정점 중심 그래프 처리 시스템을 대상으로 하여 블록 중심 그래프 처리 시스템에 적용이 어렵다. 본 논문에서는 블록 중심 그래프 처리 시스템을 대상으로 적용 가능한 로드 밸런싱 기법을 제안한다. 제안 기법은 동적으로 블록을 재배치하여 점진적으로 그래프 분할 상태를 개선시키며, 해를 찾아나가는 과정에서 지역 최적해를 벗어나기 위한 블록 분할 전략을 함께 제시한다.

• Journal of Biosystems Engineering
• /
• 제2권1호
• /
• pp.7-24
• /
• 1977

This paper is the forth and fifth one of the study on balanced type oscillating mole drainer. In the light of the results from previous reports about the model tests, some design criteria were established and a prototype machine was set up for experimental purpose. Motion characteristics and functionof the each parts of the machine were checked and analyzed. After that, performance tests of the prototype machine were carried out in thefield. Obtained results are summarized as follows ; 1. Ten centimeter of the bullet diameter was determined so as to be able to attach it to the tractors with capacity of 30 PS to 40 PS. 2. To maintain the balance between the moments of the front shank and rear shank, the oscillating amplitude of the rear bullet was determined to be larger than that of the front bullet. At the same time , the oscillating direction of the rear bullet was designed with the inclines of ten to thirty degrees. 3. An octagonal dynamo transduced was developed for measuring the compressive force of the upper link is measuring the draft force of the machine. Acceptable linear relationship between forces and strain responses from O.D.T. was obtained. 4. Analysing the balancing mechanism of the acting part of the machine , it was found that the total draft force of the machine was equal to the difference between the sum of the draft force produced from the right and left side bending moments of the lower drawber and the compressive force on the upper link. 5. There are acceptable linear relationship between the strain and twisting moment by driving shaft, and between strain and shank moment. Above results enable us to carry out the field experiment with prototype machine. 6. When the test machine was used in the field, it was possible to reduce the oscillating acceleration by forty percent in average as compared it with the single bullet mole drainer. 7. When the test machine was used under the oscillating condition, the dratt torce was reduced by 27 percent to 59 percent as compared it with the test machine under non-oscillating condition, while the draft force was increased by 7 percent to 20 percent as compared it with the mole drainer having oscillating single bullet. The reasoning behind this fact was considered as the resistance force due to the rear shank and bullet. 8. As the amplitude and frequency of the bullet were increased, the torque was increased accordingly. This tendency could be varied with the various characteristics of the given soils. And the larger frequency and amplitute, the more increasing oscil\ulcornerlating power but decreasing draft brce were needed, and draft force was increased as the velocity was increased.9. When the amplitude of the rear bullet was designed to be larger than that of the front bullet, the minimum value of the moment was lowered and oscillating acceleration was reduced. And when the oscillating direction of the rear bullet was declined back\ulcornerwards, oscillating acceleration was increased along with the increasing angle of decli\ulcornernation. When the test machine was operated in high speed, the difference between maximum moments and minimum ones became narrow. This varying magnitude of moments appeared on the moment oscillogram seems to be correlated to the oscillating acceleration and draft force. 10. From the analysis of variance, it was found that those factors such as frequency, amplitude, and operating velocity significantly affected in the oscillating acceleration, the draft resistance, the torque, the moment, and the total power required. And interaction between frequency and amplitude affected in the oscillating acceleration. 11. Within the given situation of this study, the most preferable operating conditions of the test machine were 7 Hz in oscillating frequency, 0.54 m/sec in operating velocity, and 39.1 mm in oscillating amplitude of front and rear bullets. However, it is necessary to select the proper frequency and magnitude of oscillation depending on the soil properties of the field in which the mole drainer is practiced by use of a bal1nced type oscillating mole drainer. 12. It is recommended that a comparative study of the mole drainers would be performed in the near future using two separate balanced oscillating bullet with the one which is operated by oscillating the movable bullet in a single cylinder or other balanced type which may be single oscillating bullet with spring, damper or balancing weight, and that of thing. To expand the applicability of the balanced type oscillating mole drainer in practical use, it is suggested to develop a new mechanism which perform mole drain with vinyl pipe or filling material such as rice hull.

• 한국정보통신학회논문지
• /
• 제15권3호
• /
• pp.567-575
• /
• 2011

클라우드 컴퓨팅 환경에서 가상 머신의 할당 기법인 매치메이킹, 라운드 로빈은 노드의 CPU 사용률을 고려하지 않고 CPU, RAM, HDD 등의 하드웨어 사양만으로 가상 머신을 할당한다. 이렇게 노드의 자원 할당 가능 여부만을 고려한 노드 선택 과정은 자원의 균등한 분배 측면에서 효율적이라 할 수 있지만 자원의 사용률 측면에서는 효율적이지 못하다. 따라서 본 논문에서는 사용자의 과거 가상 머신 사용률을 기반으로 노드의 CPU 사용률을 계산해 가장 작은 CPU 사용률을 갖는 노드를 선택하여 가상 머신을 할당하는 기법을 제안한다. 제안한 할당 기법의 효율성을 검증하기 위해 매치메이킹, 라운드로빈 할당 기법과 비교 실험을 진행하였고, 실험 결과 제안하는 할당 기법을 적용한 노드들이 타 기법을 적용한 노드들보다 전체적인 CPU 사용률에 있어 고른 분포를 보여 제안하는 할당 기법이 노드의 부하 분산에 효율적임을 입증하였다.