• Proceedings of the KIEE Conference
• /
• 2005.05a
• /
• pp.15-17
• /
• 2005

Recently genetic algorithm techniques have widely used in adaptive and control schemes for production systems. However, generally it costs a lot of time for learning in the case applied in control system. Furthermore, the physical meaning of genetic algorithm constructed as a result is not obvious. And this method has been used as a learning algorithm to estimate the parameter of a genetic algorithm used for identification of the process dynamics of nonlinear IIR filter and it was shown that this method offered superior capability over the genetic algorithm. A genetic algorithm is used to solve the parameter identification problem for linear and nonlinear digital filters. This paper goal estimate nonlinear IIR filter parameter using the genetic algorithm.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.5 no.3
• /
• pp.52-63
• /
• 1995

A neuro-fuzzy algorithm is presented for nuclear reactor power control in a pressurized water reactor. Automatic reacotr power control is complicated by the use of control rods because of highly nonlinear dynamics in the axial power shape. Thus, manual shaped controls are usually employed even for the limited capability during the power maneuvers. In an attempt to achieve automatic shape control, a neuro-fuzzy approach is considered because fuzzy algorithms are good at various aspects of operator's knowledge representation while neural networks are efficinet structures capable of learning from experience and adaptation to a changing nuclear core state. In the proposed neuro-fuzzy control scheme, the rule base is formulated based ona multi-input multi-output system and the dynamic back-propagation is used for learning. The neuro-fuzzy powere control algorithm has been tested using simulation fesponses of a Korean standard pressurized water reactor. The results illustrate that the proposed control algorithm would be a parctical strategy for automatic nuclear reactor power control.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.11 no.5
• /
• pp.395-399
• /
• 2001

This paper describes modular fuzzy inference systems(MFIS) with adaptive capability to extract fuzzy inference modules from observation data through the learning process. The proposed MFIS is based on the structural similarity to Tagaki-Sugeno fuzzy models and a modular neural architecture. The learning of MFIS is done by assigning new fuzzy inference modules and by updating the parameters of existing modules. The fuzzy inference modules consist of local model network and fuzzy gating network. The parameters of the MFIS are updated by the standard LMS algorithm. The performance of the MFIS is illustrated with adaptive control of a nonlinear dynamic system.

• Proceedings of the KIEE Conference
• /
• 2005.10b
• /
• pp.502-504
• /
• 2005

Recently genetic algorithm techniques have widely used in adaptive and control schemes for production systems. However, generally it costs a lot of time for learning in the case applied in control system. Furthermore, the physical meaning of genetic algorithm constructed as a result is not obvious. And this method has been used as a learning algorithm to estimate the parameter of a genetic algorithm used for identification of the process dynamics of FIR filter and it was shown that this method offered superior capability over the genetic algorithm. A genetic algorithm is used to solve the parameter identification problem for linear and nonlinear digital filters. This paper goal estimate FIR filter parameter using the genetic algorithm.

• Proceedings of the KIEE Conference
• /
• 1992.07a
• /
• pp.484-487
• /
• 1992

To resolve the knowledge acquisition bottleneck in the expert systems, the connectionist expert systems have been proposed, which facilitate learning capability of neural networks. This paper is to modify Gallant's connectionist expert network so that it can be applied to more general problems : 1) The hidden nodes are added between the input nodes and an output node, so that the back propagation learning algorithm is used instead of perception based Pocket algorithm. 2) Inference engine is thus modified by modeling that a node may have uncertainties due to unknown inputs.

• Journal of the Korean Institute of Telematics and Electronics C
• /
• v.34C no.9
• /
• pp.61-67
• /
• 1997

The HMM-Net is an architecture for a neural network that implements a hidden markov model(HMM). The architecture is developed for the purpose of combining the classification power of neural networks with the time-domain modeling capability of HMMs. Criteria which are used for learning HMM_Net classifiers are maximum likelihood(ML), maximum mutual information (MMI), and minimization of mean squared error(MMSE). In this classifiers trained by the gradient descent algorithm with the above criteria. Experimental results for the isolated numbers from /young/to/koo/ show that in the binary inputs the performance of MMSE is better than the others, while in the fuzzy inputs the performance of MMI is better than the others.

• Proceedings of the Korean Operations and Management Science Society Conference
• /
• 1996.04a
• /
• pp.244-248
• /
• 1996

Nowadays, most shop floors using industrial robots have many problems such as constructing robot workcell, generating robot arm moving trajectory, etc.. In the case of programming robot-arms for a specific task, shop operator commonly use the teach pendant to record the target position and determine the moving trajectory. However, such a teaching process may result in an inefficient trajectory in the sense of moving distance and joint angle fluctuation. Moreover, shop operators who have little knowledge about robot programming process need a lot of learning time and cost. The purpose of this paper is to propose a user friendly robot programming method to program robot-arms easily and efficiently for shop operator so that the programming time is reduced and a short and stable trajectory is obtained.

• 제어로봇시스템학회:학술대회논문집
• /
• 1992.10b
• /
• pp.61-64
• /
• 1992

In this paper, a visual pattern recognition system is proposed, which can recognize both a pattern and its location. This system, referred to as the expanded neocognitron, has the following capabilities: (1) A higher performance in extraction of features, and (2) A new capability for recognizing the locations of patterns. This system adopts the learning and recognizing mechanism of the neocognitron. First, the ability to classify pattern is enhanced by improving the mechanisms of feature extraction and learning algorithm. Second, the function of detecting the location of each pattern is realized by developing an architecture which does not reduce structure, i.e., the unit density is constant all the way from the input stage to the output stage.

• Proceedings of the IEEK Conference
• /
• 2003.07e
• /
• pp.2016-2019
• /
• 2003

In this paper, we take some features for face recognition out of face image, using a simple type of templates. We use the extracted features to do Adaboost learning for face recognition. Using a carefully-chosen feature among these features, we can make a weak face classifier for face recognition. And doing Adaboost learning on and on with those chosen several weak classifiers, we can get a strong face classifier. By using Adaboost Loaming, we can choose particular features which is not easily subject to changes in illumination and facial expression about several images of one person, and construct face recognition system. Therefore, the face classifier bulit like the above way has robustness in both facial expression and illumination variation, and it finally gives capability of recognizing face fast due to the simple feature.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.34 no.3C
• /
• pp.257-267
• /
• 2009

This paper presents a training method for neural networks and the employment of MSE (mean scare error) values as the basis of a decision regarding the identity claim of a speaker in a recurrent neural networks based speaker verification system. Recurrent neural networks (RNNs) are employed to capture temporally dynamic characteristics of speech signal. In the process of supervised learning for RNNs, target outputs are automatically generated and the generated target outputs are made to represent the temporal variation of input speech sounds. To increase the capability of discriminating between the true speaker and an impostor, a discriminative training method for RNNs is presented. This paper shows the use and the effectiveness of the MSE value, which is obtained from the Euclidean distance between the target outputs and the outputs of networks for test speech sounds of a speaker, as the basis of speaker verification. In terms of equal error rates, results of experiments, which have been performed using the Korean speech database, show that the proposed speaker verification system exhibits better performance than a conventional hidden Markov model based speaker verification system.