• Journal of the Semiconductor & Display Technology
• /
• v.20 no.1
• /
• pp.64-69
• /
• 2021

In this paper, a method for improving the defect classification performance in low contrast, ununiformity and featureless steel plate surfaces has been studied based on deep convolution neural network and transfer-learning neural network. The steel plate surface images have low contrast, ununiformity, and featureless, so that the contrast between defect and defect-free regions are not discriminated. These characteristics make it difficult to extract the feature of the surface defect image. A classifier based on a deep convolution neural network is constructed to extract features automatically for effective classification of images with these characteristics. As results of the experiment, AlexNet-based transfer-learning classifier showed excellent classification performance of 99.43% with less than 160 seconds of training time. The proposed classification system showed excellent classification performance for low contrast, ununiformity, and featureless surface images.

• Journal of Information Processing Systems
• /
• v.17 no.3
• /
• pp.556-570
• /
• 2021

The existing video expression recognition methods mainly focus on the spatial feature extraction of video expression images, but tend to ignore the dynamic features of video sequences. To solve this problem, a multi-mode convolution neural network method is proposed to effectively improve the performance of facial expression recognition in video. Firstly, OpenFace 2.0 is used to detect face images in video, and two deep convolution neural networks are used to extract spatiotemporal expression features. Furthermore, spatial convolution neural network is used to extract the spatial information features of each static expression image, and the dynamic information feature is extracted from the optical flow information of multiple expression images based on temporal convolution neural network. Then, the spatiotemporal features learned by the two deep convolution neural networks are fused by multiplication. Finally, the fused features are input into support vector machine to realize the facial expression classification. Experimental results show that the recognition accuracy of the proposed method can reach 64.57% and 60.89%, respectively on RML and Baum-ls datasets. It is better than that of other contrast methods.

• The Transactions of the Korea Information Processing Society
• /
• v.3 no.5
• /
• pp.1138-1146
• /
• 1996

Fuzzy logic ignores some informations in the reasoning process. Neural networks are powerful tools for the pattern processing. However, to model human knowledges, besides pattern processing capability, the logical reasoning capability is equally important. Another new neural network called neural logic network is able to do the logical reasoning. Because the fuzzy logical reasoning, we construct fuzzy inference net-work based on the neural logic network, extending the existing rule-inferencing network. And the traditional propagation rule is modified. For the search strategies to find out the belief value of a conclusion in the fuzzy inference network, we conduct a simulation to evaluate the search cost for searching sequentially and searching by means of priorities.

• ETRI Journal
• /
• v.41 no.6
• /
• pp.760-770
• /
• 2019

Based on the growing demand for neural network technologies, various neural network inference engines are being developed. However, each inference engine has its own neural network storage format. There is a growing demand for standardization to solve this problem. This study presents interworking techniques for ensuring the compatibility of neural networks and data among the various deep learning frameworks. The proposed technique standardizes the graphic expression grammar and learning data storage format using the Neural Network Exchange Format (NNEF) of Khronos. The proposed converter includes a lexical, syntax, and parser. This NNEF parser converts neural network information into a parsing tree and quantizes data. To validate the proposed system, we verified that MNIST is immediately executed by importing AlexNet's neural network and learned data. Therefore, this study contributes an efficient design technique for a converter that can execute a neural network and learned data in various frameworks regardless of the storage format of each framework.

• Journal of Korea Society of Digital Industry and Information Management
• /
• v.19 no.1
• /
• pp.109-118
• /
• 2023

Data-based analysis methods have become used more for estimating or predicting housing prices, and neural network models and decision trees in the field of big data are also widely used more and more. Neural network models are often evaluated to be superior to existing statistical models in terms of estimation or prediction accuracy. However, there is ambiguity in determining the input feature of the input layer of the neural network model, that is, the type and number of input features, and decision trees are sometimes used to overcome these disadvantages. In this paper, we evaluate the existing methods of using decision trees and propose the method of using decision trees to prioritize input feature selection in neural network models. This can be a complementary or combined analysis method of the neural network model and decision tree, and the validity was confirmed by applying the proposed method to house price estimation. Through several comparisons, it has been summarized that the selection of appropriate input characteristics according to priority can increase the estimation power of the model.

• Proceedings of the IEEK Conference
• /
• 2000.07b
• /
• pp.1053-1056
• /
• 2000

In general, neural networks based modeling involves trying multiple networks with different architectures and/or training parameters in order to achieve the best accuracy. Only the single best-trained neural network is chosen, while the rest are discarded. However, using only the single best network may never give the best solution in every situation. Many researchers, therefore, propose methods to improve the accuracy of neural networks based modeling. In this paper, the idea of the logical combinations of neural networks is proposed and discussed in detail. The logical combination is constructed by combining the corresponding outputs of the neural networks with the logical “And” node. The experimental results based on simulated data show that the modeling accuracy is significantly improved when compared to using only the single best-trained neural network.

• 제어로봇시스템학회:학술대회논문집
• /
• 1992.10a
• /
• pp.243-247
• /
• 1992

This paper describes a neural network based control scheme with MRAC. The system consists of two neural network; one is for identifier and the other is for controller. Identification is firstly performed to learn the behavior of the nonlinear plant. Neural net controller is next trained by backpropagating the error at the output of plant through the identifier. Also the training method used in this paper repeatedly updates weights of neural network to track the reference model.

• 제어로봇시스템학회:학술대회논문집
• /
• 2000.10a
• /
• pp.74-74
• /
• 2000

This paper presents a neural network controller for a rigid-link electrically-driven robot. The proposed controller is designed in conjunction with three neural networks approximating for complicated nonlinear functions. Particularly, the fact, different from conventional schemes, is that the neural network based current observer is used. Therefore, no accurate measurement of the actuator driving current is required. In the proposed controller-observer scheme, the derived weight update rule guarantees the stability of closed-loop system in the sense of Lyapunov. The effectiveness and performance of the proposed method are demonstrated through computer simulation.

• 제어로봇시스템학회:학술대회논문집
• /
• 2005.06a
• /
• pp.1716-1722
• /
• 2005

Fuzzy logic, neural network, fuzzy-neural network play an important as the key technology of linguistic modeling for intelligent control and decision in complex systems. The fuzzy-neural network (FNN) learning represents one of the most effective algorithms to build such linguistic models. This paper proposes bacteria foraging algorithm based optimal learning fuzzy-neural network (BA-FNN). The proposed learning scheme is the fuzzy-neural network structure which can handle linguistic knowledge as tuning membership function of fuzzy logic by bacteria foraging algorithm. The learning algorithm of the BA-FNN is composed of two phases. The first phase is to find the initial membership functions of the fuzzy neural network model. In the second phase, bacteria foraging algorithm is used for tuning of membership functions of the proposed model.

• Transactions of the Korean Society of Automotive Engineers
• /
• v.10 no.2
• /
• pp.132-140
• /
• 2002

A momentum backpropagation neural network is prepared to carry out real-time dynamics simulations of a passenger car. A full-car model of fifteen degrees of freedom was constructed for vehicle dynamics analysis. Human body dynamics analysis was performed for a male driver(50 percentile Korean adult) restrained by a three point seatbelt system. The trained data using the neural network were obtained using a dynamic solver, ADAMS . The neural network were formed based on the dynamics of the simulator. The optimized hidden layer was obtained by selecting the optimal number of hidden layers. The driving scenario including bump passing and lane changing has been used for the estimation of the proposed neural network. A comparison between the trained data and neural network outputs is found to be satisfactory to show the applicability of the suggested approach.