• Proceedings of the Korean Institute of Intelligent Systems Conference
• /
• 2003.05a
• /
• pp.85-88
• /
• 2003

강화학습의 한가지 방법인 Q-learning은 최근에 Linear Quadratic Regulation(이하 LQR) 문제에 성공적으로 적용된 바 있다. 특히, 시스템 모델의 파라미터에 대한 구체적인 정보없이 적절한 입ㆍ출력만으로 학습을 통해 문제의 해결이 가능하므로 상황에 따라 매우 실용적인 방법이 될 수 있다. 뉴럴-큐 기법은 이러한 Q-learning의 Q-value를 MLP(multilayer perceptron) 신경망의 출력으로 대치시켜, 비선형 시스템의 최적제어 문제를 다룰 수 있게 한 방법이다. 그러나, 뉴럴-큐 기법은 신경망의 구조를 먼저 결정한 후 역전파 알고리즘을 이용해 학습하는 절차를 행하므로, 시행착오를 통해 신경망 구조를 결정해야 한다는 점, 역전파 알고리즘의 적용에 따라 신경망의 연결강도 값들이 지역적 최적해로 수렴한다는 점등의 문제점이 있다. 본 논문에서는 뉴럴-큐 학습의 도구로 KFD회귀를 이용하여 Q 함수의 근사 기법을 제안하고 관련 수식을 유도하였다. 그리고, 모의 실험을 통하여, 제안된 뉴럴-큐 방법의 적용 가능성을 알아보았다.

• International Journal of Fuzzy Logic and Intelligent Systems
• /
• v.2 no.3
• /
• pp.174-178
• /
• 2002

This paper discusses a blind equalization technique for FIR channel system, that might be minimum phase or not, in digital communication. The proposed techniques consist of two parts. One is to estimate the original channel coefficients based on fourth-order cumulants of the channel output, the other is to employ RBF neural network to model an inverse system fur the original channel. Here, the estimated channel is used as a reference system to train the RBF. The proposed RBF equalizer provides fast and easy teaming, due to the structural efficiency and excellent recognition-capability of R3F neural network. Throughout the simulation studies, it was found that the proposed blind RBF equalizer performed favorably better than the blind MLP equalizer, while requiring the relatively smaller computation steps in tranining.

• Journal of Biomedical Engineering Research
• /
• v.15 no.4
• /
• pp.455-464
• /
• 1994

The intelligent trajectory control method that controls moving direction and average velocity for a prosthetic arm is proposed by pattern recognition and force estimations using EMG signals. Also, we propose the real time trajectory planning method which generates continuous accelleration paths using 3 stage linear filters to minimize the impact to human body induced by arm motions and to reduce the muscle fatigue. We use combination of MLP and fuzzy filter for pattern recognition to estimate the direction of a muscle and Hogan's method for the force estimation. EMG signals are acquired by using a amputation simulator and 2 dimensional joystick motion. The simulation results of proposed prosthetic arm control system using the EMG signals show that the arm is effectively followed the desired trajectory depended on estimated force and direction of muscle movements.

• 제어로봇시스템학회:학술대회논문집
• /
• 2000.10a
• /
• pp.431-431
• /
• 2000

The modeling of 5-bar linkage robot manipulator dynamics by means of a mathematical and neural architecture is presented. Such a model is applicable to the design of a feedforward controller or adjustment of controller parameters. The inverse model consists of two parts: a mathematical part and a compensation part. In the mathematical part, the subsystems of a 5-bar linkage robot manipulator are constructed by applying Kawato's Feedback-Error-Learning method, and trained by given training data. In the compensation part, MLP backpropagation algorithm is used to compensate the unmodeled dynamics. The forward model is realized from the inverse model using the inverse of inertia matrix and the compensation torque is decoupled in the input torque of the forward model. This scheme can use tile mathematical knowledge of the robot manipulator and analogize the robot characteristics. It is shown that the model is reasonable to be used for design and initial gain tuning of a controller.

• Proceedings of the Korean Society of Precision Engineering Conference
• /
• 1994.10a
• /
• pp.510-514
• /
• 1994

This paper deals with an application of neural network to camera calibration with wide angle lens and 2-D range finding. Wide angle lens has an advantage of having wide view angles for mobile environment recognition ans robot eye in hand system. But, it has severe radial distortion. Multilayer neural network is used for the calibration of the camera considering lens distortion, and is trained it by error back-propagation method. MLP can map between camera image plane and plane the made by structured light. In experiments, Calibration of camers was executed with calibration chart which was printed by using laser printer with 300 d.p.i. resolution. High distortion lens, COSMICAR 4.2mm, was used to see whether the neural network could effectively calibrate camera distortion. 2-D range of several objects well be measured with laser range finding system composed of camera, frame grabber and laser structured light. The performance of 3-D range finding system was evaluated through experiments and analysis of the results.

• Journal of Korea Society of Industrial Information Systems
• /
• v.16 no.2
• /
• pp.31-38
• /
• 2011

Skin color is a very important information for an automatic face recognition. In this paper, we proposed a skin region extraction method using the MLP(Multi-Layer Perceptron) and skin color model. We use the adaptive lighting compensation technique for improved performance of skin region extraction. Also, using an preprocessing filter, normally large areas of easily distinct non-skin pixels, are eliminated from further processing. Experimental results show that the proposed method has better performance than the conventional methods, and reduces processing time by 31~49% on average.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.28 no.7C
• /
• pp.705-711
• /
• 2003

This paper studies the application of a fuzzy-ARTMAP neural network to digital communications channel equalization. This approach provides new solutions for solving the problems, such as complexity and long training, which found when implementing the previously developed neural-basis equalizers. The proposed fuzzy-ARTMAP equalizer is fast and easy to train and includes capabilities not found in other neural network approaches; a small number of parameters, no requirements for the choice of initial weights, automatic increase of hidden units, no risk of getting trapped in local minima, and the capability of adding new data without retraining previously trained data. In simulation studies, binary signals were generated at random in a linear channel with Gaussian noise. The performance of the proposed equalizer is compared with other neural net basis equalizers, specifically MLP and RBF equalizers.

• Journal of the Korean Institute of Telematics and Electronics B
• /
• v.28B no.5
• /
• pp.382-393
• /
• 1991

In this paper, we propose a method for recognizing printed Korean characters using the Preceding Layer Driven multi-layer perceptron. The new learning algorithm which assigns the weight values to an integer and makes use of the transfer function as the step function was presented to design the hardware. We obtained 522 Korean character-image as an experimental object through scanner with 600DPI resolution. The preprocessing for feature extraction of Korean character is the separation of individual character, noise elimination smoothing, thinnig, edge point extraction, branch point extraction, and stroke segmentation. The used feature data are the number of edge points and their shapes, the number of branch points, and the number of strokes with 8 directions.

• Journal of the Semiconductor & Display Technology
• /
• v.19 no.3
• /
• pp.77-81
• /
• 2020

In this paper, we create scaffold defect classification models using machine learning based data. We extract the characteristic from collected scaffold external images using USB camera. SVM, KNN, MLP algorithm of machine learning was using extracted features. Classification models of three type learned using train dataset. We created scaffold defect classification models using test dataset. We quantified the performance of defect classification models. We have confirmed that the SVM accuracy is 95%. So the best performance model is using SVM.

• Proceedings of the Computational Structural Engineering Institute Conference
• /
• 2011.04a
• /
• pp.53-56
• /
• 2011

본 논문에서는 지진시 구조물의 진동을 줄이기 위한 방법으로 모드에너지 기반 신경망 제어 방법을 제안하였다. 모드에너지 기반 신경망 제어 방법은 신경망의 학습 과정에서 구조물의 모드 에너지를 이용하여 목적함수를 구성하며, 이 목적함수를 최소로 하는 학습을 진행한다. 제안된 제어 알고리즘의 적용성을 검증하기 위해서 능동질량감쇠기(AMD, Active Mass Damper)가 설치된 3층 구조물을 예제 모델로 선택하였으며, El Centrol지진을 이용하여 모드에너지기반 신경망제어 알고리즘을 학습시켰다. 모드에너지 기반 신경망 제어 알고리즘의 제어 성능은 학습 후 임의의 지진에 대한 하중으로 California지진을 사용하여 검증하였다. 해석 결과에서 California지진에 대한 제어 전 후의 결과와 기존의 방법인 MLP(Muli-layer Perceptron)의 결과와 비교하였다. 또한 제안된 제어 방법을 적용할 때, 지진시 구조물의 비선형 거동은 제어후 거의 보이지 않는 것을 확인 할 수 있었다.