• 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.

• 제어로봇시스템학회:학술대회논문집
• /
• 2000.10a
• /
• pp.229-229
• /
• 2000

Conventional gain-tuning methods such as Ziegler-Nickels methods, have many disadvantages that optimal control ler gain should be tuned manually. In this paper, modified PID controllers which include self-tuning characteristics are proposed. Proposed controllers automatically tune the PID gains in on-1ine using neural networks. A new learning scheme was proposed for improving learning speed in neural networks and satisfying the real time condition. In this paper, using a nonlinear mapping capability of neural networks, we derive a tuning method of PID controller based on a Back propagation(BP)method of multilayered neural networks. Simulated and experimental results show that the proposed method can give the appropriate parameters of PID controller when it is implemented to DC Motor.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.23 no.9A
• /
• pp.2211-2220
• /
• 1998

This paper presetns an adaptive critic self-learning control system with cerebellar model articulation controller (CMAC)-based decoder integrated with the associative search element (ASE) and adatpive critic element(ACE)- based scheme. The tast of the system is to balance a pole that is hinged to a movable cart by applying forces to the cart's base. The problem is that error feedback information is limited. This problem can be sloved when some adaptive control devices are involved. The ASE incorporates prediction information for reinforrcement from a critic to produce evaluative information for the plant. The CMAC-based decoder interprets one state to a set of patways into the ASE/ACE. These signals correspond to te current state and its possible preceding action states. The CMAC's information interpolation improves the learning speed. And design inverted pendulum hardware system to show control capability with neural network.

• International Journal of Fuzzy Logic and Intelligent Systems
• /
• v.5 no.3
• /
• pp.263-267
• /
• 2005

Building classifiers for financial real-world classification problems is often plagued by severely overlapping and highly skewed class distribution. New performance measures such as receiver operating characteristic (ROC) curve and area under ROC curve (AUC) have been recently introduced in evaluating and building classifiers for those kind of problems. They are, however, in-effective to evaluation of classifier's discrimination performance in a particular class of the classification problems that interests lie in only a local operating range of the classifier, In this paper, a new method is proposed that enables us to directly improve classifier's discrimination performance at a desired local operating range by defining and optimizing a partial area under ROC curve or domain-specific curve, which is difficult to achieve with conventional classification accuracy based learning methods. The effectiveness of the proposed approach is demonstrated in terms of fraud detection capability in a real-world fraud detection problem compared with the MSE-based approach.

• Journal of Intelligence and Information Systems
• /
• v.1 no.2
• /
• pp.57-71
• /
• 1995

This paper presents a rule extraction algorithm RE to acquire explicit rules from trained neural networks. The validity of extracted rules has been confirmed using 6 different data sets. Based on experimental results, we conclude that extracted rules from RE predict more accurately and robustly than neural networks themselves and rules obtained from an inductive learning algorithm do. Rule extraction algorithm for neural networks are important for incorporating knowledge obtained from trained networks into knowledge based systems. In lieu of this, the proposed RE algorithm contributes to the trend toward developing hybrid and versatile knowledge-based system including expert systems and knowledge-based decision su, pp.rt systems.

• Journal of Educational Research in Mathematics
• /
• v.12 no.4
• /
• pp.543-556
• /
• 2002

The purpose of this study is to find a method to improve geometry instruction. For this purpose, I have investigated aims and problems of geometry education. I also reviewed related literature about discovery methods as well as verification. Through this review, “Mathematising teaching and learning methods” by Freudenthal is Presented as an alternative to geometry instruction. I investigated the capability of dynamic software for realization of this method. The result of this investigation is that dynamic software is a powerful tool in realizing this method. At last, I present one example of mathematic activity using dynamic software that can be used by school teachers.

• Proceedings of the Optical Society of Korea Conference
• /
• 1989.02a
• /
• pp.85-88
• /
• 1989

Introducing and optimizing it-significance to the Hopfield model, ten highly correlated binary images, i.e., numbers "0" to "9", are successfully stored and retrieved in a 6x8 node system. Unlike many other neural networks models, this model has stronger error correction capability for correlated images such as "6", "8", "3", and "9". the bit-significance optimization is regarded as an adaptive learning process based on least-mean-square error algorithm, and may be implemented with another neural nets optimizer. A design for electro-optic implementation including the adaptive optimization networks is also introduced.uding the adaptive optimization networks is also introduced.

• Journal of the Korean Chemical Society
• /
• v.65 no.2
• /
• pp.145-150
• /
• 2021

The present study describes a hands-on experiment to help students understand the concept of phase change or phase transition and its application in a phase change material (PCM). PCMs are substances that have the capability of storing and releasing large amounts of thermal energy. They act as energy storage materials that provide an effective way to save energy by reducing the electricity required for heating and cooling. Lauric acid (LA) was selected as an example of the PCM. Students investigated the temperature change of LA and the temperature (of air) inside the test tube. The differences in the temperatures of the systems helped students understand how PCMs work. A one-group pretest and posttest design was implemented with 34 grade-11 students in science and mathematics. Students' understanding was assessed using a multiple-choice test and a questionnaire. The findings revealed that the designed activity helped students understand the concept of phase change and its application to materials for thermal energy storage.