• 제어로봇시스템학회:학술대회논문집
• /
• 1998.10a
• /
• pp.250-253
• /
• 1998

This paper presents a multi-modal neural network composed of a preprocessing module and a multi-layer neural network module in order to enhance the nonlinear characteristics of neural network. The former module is based on spectral method using Chebyschev polynomials and transforms input data into spectra. The latter module identifies the system using the spectra generated by the preprocessing module. The omnibus numerical experiments show that the method is applicable to many a nonlinear dynamic system in the real world, and that preprocessing using Chebyschev polynomials reduces the number of neurons required for the multi-layer neural network.

• Journal of the Institute of Convergence Signal Processing
• /
• v.2 no.2
• /
• pp.111-118
• /
• 2001

In this paper, in order to implement the multi-layered perceptron neural network using pure digital logic circuit model, we propose the new logic neuron structure, the digital canonical multi-layered logic neural network structure, and the multi-stage multi-layered logic neural network structure for pattern recognition applications. And we show that the proposed approach provides an incremental additive learning algorithm, which is very simple and effective.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.22 no.1
• /
• pp.40-48
• /
• 2018

This paper proposes a license-plate recognition algorithm for automobile black-box image which is obtained from the camera moving with the automobile. The algorithm intends to increase the overall recognition-rate of the license-plate by increasing the Korean character recognition-rate using multi-stage neural network for automobile black-box image where there are many movements of the camera and variations of light intensity. The proposed algorithm separately recognizes the vowel and consonant of Korean characters of automobile license-plate. First, the first-stage neural network recognizes the vowels, and the recognized vowels are classified as vertical-vowels('ㅏ','ㅓ') and horizontal-vowels('ㅗ','ㅜ'). Then the consonant is classified by the second-stage neural networks for each vowel group. The simulation for automobile license-plate recognition is performed for the image obtained by a real black-box system, and the simulation results show the proposed algorithm provides the higher recognition-rate than the existing algorithms using a neural network.

• Journal of the Korean Institute of Telematics and Electronics B
• /
• v.31B no.1
• /
• pp.88-98
• /
• 1994

A new neural network architecture called the Parallel. Self-Organizing Hierarchical Neural Network (PSHNN) is presented. The new architecture involves a number of stages in which each stage can be a particular neural network (SNN). The experiments performed in comparison to multi-layered network with backpropagation training and indicated the superiority of the new architecture in the sense of classification accuracy, training time,parallelism.

• Journal of Korea Water Resources Association
• /
• v.50 no.2
• /
• pp.111-119
• /
• 2017

The underflow type movable weirs were arranged in a multi-stage way along a reach at the Chiseong River, where flooding has been observed frequently. With management water level of the movable weirs the control effects of storage and flood were suggested and the control effects were compared with those of existed weir system. The water level for the targeted storage and flood elevation was suggested by building the artificial neural network model. When the underflow type of movable weirs were arranged in a multi-stage way, the peak flood elevation decreased by 68.28% in the downstream compared with the existed weir system, and the total storage of the target section of multi-stage movable weirs increased by 216%. As a result of numerical simulation to build the artificial neural network model, 60%, 20%, and 20% among 216 data were used for the training, validation, and test, respectively. The training result of mean square error was $0.1681m^2$ and the high coefficients of determination were 0.9961, 0.9967, and 0.9943 in the training, validation, and test, respectively. As a result the water level management of each movable weir for the controls of flood elevation in the targeted downstream and targeted storage was suggested by using the artificial neural network.

• The korean journal of orthodontics
• /
• v.51 no.2
• /
• pp.77-85
• /
• 2021

Objective: To evaluate the accuracy of a multi-stage convolutional neural network (CNN) model-based automated identification system for posteroanterior (PA) cephalometric landmarks. Methods: The multi-stage CNN model was implemented with a personal computer. A total of 430 PA-cephalograms synthesized from cone-beam computed tomography scans (CBCT-PA) were selected as samples. Twenty-three landmarks used for Tweemac analysis were manually identified on all CBCT-PA images by a single examiner. Intra-examiner reproducibility was confirmed by repeating the identification on 85 randomly selected images, which were subsequently set as test data, with a two-week interval before training. For initial learning stage of the multi-stage CNN model, the data from 345 of 430 CBCT-PA images were used, after which the multi-stage CNN model was tested with previous 85 images. The first manual identification on these 85 images was set as a truth ground. The mean radial error (MRE) and successful detection rate (SDR) were calculated to evaluate the errors in manual identification and artificial intelligence (AI) prediction. Results: The AI showed an average MRE of 2.23 ± 2.02 mm with an SDR of 60.88% for errors of 2 mm or lower. However, in a comparison of the repetitive task, the AI predicted landmarks at the same position, while the MRE for the repeated manual identification was 1.31 ± 0.94 mm. Conclusions: Automated identification for CBCT-synthesized PA cephalometric landmarks did not sufficiently achieve the clinically favorable error range of less than 2 mm. However, AI landmark identification on PA cephalograms showed better consistency than manual identification.

• The Journal of the Acoustical Society of Korea
• /
• v.36 no.6
• /
• pp.425-435
• /
• 2017

In this paper, we propose optimization methods for speeding the feedforward network of deep neural networks using NEON SIMD (Single Instruction Multiple Data) parallel instructions and multi-core parallelization on the multi-core ARM processor. As the result of the optimization using SIMD parallel instructions, we present the amount of speed improvement and arithmetic precision stage by stage. Through the optimization using SIMD parallel instructions on the single core, we obtain $2.6{\times}$ speedup over the baseline implementation using C compiler. Furthermore, by parallelizing the single core implementation on the multi-core, we obtain $5.7{\times}{\sim}7.7{\times}$ speedup. The results we obtain show the possibility for applying the arithmetic-intensive deep neural network technology to applications on mobile devices.

• Transactions of Materials Processing
• /
• v.29 no.5
• /
• pp.272-281
• /
• 2020

As the thickness of mobile communication devices is getting thinner, the size of the internal parts is also getting smaller. Among them, the battery case requires a high-level deep drawing technique because it has a rectangular shape with a large aspect ratio. In this study, the initial blank shape was optimized to minimize earing in a multi-stage deep drawing process using an artificial neural network(ANN). There has been no reported case of applying artificial neural network technology to the initial blank optimal design for a square case with large aspect ratio. The training data for ANN were obtained though simulation, and the model reliability was verified by performing comparative study with regression model using random sample test and goodness-of-fit test. Finally, the optimal design of the initial blank shape was performed through the verified ANN model.

• Korean Journal of Remote Sensing
• /
• v.37 no.4
• /
• pp.719-731
• /
• 2021

This study proposes a two-stage hybrid classification model for crop classification using multi-temporal remote sensing images; the model combines feature embedding by using an autoencoder (AE) with a convolutional neural network (CNN) classifier to fully utilize features including informative temporal and spatial signatures. Long short-term memory (LSTM)-based AE (LAE) is fine-tuned using class label information to extract latent features that contain less noise and useful temporal signatures. The CNN classifier is then applied to effectively account for the spatial characteristics of the extracted latent features. A crop classification experiment with multi-temporal unmanned aerial vehicle images is conducted to illustrate the potential application of the proposed hybrid model. The classification performance of the proposed model is compared with various combinations of conventional deep learning models (CNN, LSTM, and convolutional LSTM) and different inputs (original multi-temporal images and features from stacked AE). From the crop classification experiment, the best classification accuracy was achieved by the proposed model that utilized the latent features by fine-tuned LAE as input for the CNN classifier. The latent features that contain useful temporal signatures and are less noisy could increase the class separability between crops with similar spectral signatures, thereby leading to superior classification accuracy. The experimental results demonstrate the importance of effective feature extraction and the potential of the proposed classification model for crop classification using multi-temporal remote sensing images.

• Nuclear Engineering and Technology
• /
• v.31 no.2
• /
• pp.192-201
• /
• 1999

This paper proposes an improved signal processing strategy for accurate feedwater flowrate estimation in nuclear power plants. It is generally known that ∼2% thermal power errors occur due to fouling Phenomena in feedwater flowmeters. In the strategy Proposed, the noises included in feedwater flowrate signal are classified into rapidly varying noises and gradually varying noises according to the characteristics in a frequency domain. The estimation precision is enhanced by introducing a low pass filter with the wavelet analysis against rapidly varying noises, and an autoassociative neural network which takes charge of the correction of only gradually varying noises. The modified multivariate stratification sampling using the concept of time stratification and MAXIMIN criteria is developed to overcome the shortcoming of a general random sampling. In addition the multi-stage robust training method is developed to increase the quality and reliability of training signals. Some validations using the simulated data from a micro-simulator were carried out. In the validation tests, the proposed methodology removed both rapidly varying noises and gradually varying noises respectively in each de-noising step, and 5.54% root mean square errors of initial noisy signals were decreased to 0.674% after de-noising. These results indicate that it is possible to estimate the reactor thermal power more elaborately by adopting this strategy.