• The Journal of Information Technology
• /
• v.7 no.2
• /
• pp.43-54
• /
• 2004

Since Dr. John J. Hopfield has proposed the HOpfield network, it has been widely applied to the pattern recognition and the routing optimization. The method of Jian-Hua Li improved efficiency of Hopfield network which input pattern's weights are regenerated by SVD(singluar value decomposition). This paper deals with Li's Hopfield Network by linear pre-processing. Linear pre-processing is used for increasing orthogonality of input pattern set. Two methods of pre-processing are used, Hadamard method and random method. In manner of success rate, radom method improves maximum 30 percent than the original and hadamard method improves maximum 15 percent. In manner of success time, random method decreases maximum 5 iterations and hadamard method decreases maximum 2.5 iterations.

• Journal of the Korean Society of Surveying, Geodesy, Photogrammetry and Cartography
• /
• v.31 no.6_2
• /
• pp.559-565
• /
• 2013

Spatial resolution of land covers from remotely sensed images can be increased using super-resolution mapping techniques for soft-classified land cover proportions. A further development of super-resolution mapping technique is downscaling the original remotely sensed image using super-resolution mapping techniques with a forward model. In this paper, the model for increasing spatial resolution of remote sensing multispectral image is tested with real SPOT 5 imagery at 10m spatial resolution for an area in Bac Giang Province, Vietnam in order to evaluate the feasibility of application of this model to the real imagery. The soft-classified land cover proportions obtained using a fuzzy c-means classification are then used as input data for a Hopfield neural network (HNN) to predict the multispectral images at sub-pixel spatial resolution. The 10m SPOT multispectral image was improved to 5m, 3,3m and 2.5m and compared with SPOT Panchromatic image at 2.5m resolution for assessment.Visually, the resulted image is compared with a SPOT 5 panchromatic image acquired at the same time with the multispectral data. The predicted image is apparently sharper than the original coarse spatial resolution image.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.14 no.2
• /
• pp.855-870
• /
• 2020

Windows are key components of building facades. Detecting windows, crucial to 3D semantic reconstruction and scene parsing, is a challenging task in computer vision. Early methods try to solve window detection by using hand-crafted features and traditional classifiers. However, these methods are unable to handle the diversity of window instances in real scenes and suffer from heavy computational costs. Recently, convolutional neural networks based object detection algorithms attract much attention due to their good performances. Unfortunately, directly training them for challenging window detection cannot achieve satisfying results. In this paper, we propose an approach for window detection. It involves an improved Faster R-CNN architecture for window detection, featuring in a window region proposal network, an RoI feature fusion and a context enhancement module. Besides, a post optimization process is designed by the regular distribution of windows to refine detection results obtained by the improved deep architecture. Furthermore, we present a newly collected dataset which is the largest one for window detection in real street scenes to date. Experimental results on both existing datasets and the new dataset show that the proposed method has outstanding performance.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.7 no.4
• /
• pp.74-81
• /
• 1997

This paper proposes an evolutionary learning algorithm to discipline the projection neural nctworks (PNNs) with special type of hidden nodes which can activate radial basis functions as well as sigmoid functions. The proposed algorithm not only trains the parameters and the connection weights hut also c~ptimizes the network structure. Through the structure optimization, the number of hidden node:; necessary to represent a given target function is determined and the role of each hidden node is decided whether it activates a radial basis function or a sigmoid function. To apply the algorithm, PNN is realized by a self-organizing genotype representation with a linked list data structure. Simulations show that the algorithm can build the PNN with less hidden nodes than thc existing learning algorithm using error hack propagation(EE3P) and network growing strategy.

• The Transactions of the Korean Institute of Power Electronics
• /
• v.6 no.6
• /
• pp.506-513
• /
• 2001

The torque of an SRM depends on the phase current and derivative of inductance. But an SRM is difficult to control the desired torques because of saturation in magnetic circuit An SRM is controlled by parameters of input voltage, and switch on , off angle The switch on off angles of an SRM regulate the magnitude and shape of current waveform and decide the magnitude and shape of torque This paper proposes an the optimization control scheme by adjusting both the switch on an switch off angle . The switch off angles are decided by reference of efficiency using simulation and experiments. The switch on angles are decided by load torque , And the dwell angles are controlled for torque control and speed control using GA-neural network which is used to simulated the reasonable switching angle.

• Journal of the Korean Institute of Telematics and Electronics C
• /
• v.36C no.10
• /
• pp.66-74
• /
• 1999

In this paper, we propose a generalized model of statistical Hopfield neural network applicable to solving the well known NP-Complete problem, TSP. Van Den Bout's method to simplify the energy function through normalization has severe weak points that it does not consider the necessary perturbation effects. In proposed model, the improved energy function is used and 5 kinds of perturbation effects and the ratio between perturbation effects are considered including van Den Bout's 2 kinds and one more kind of Park. Through the simulation of randomly generated distribution of 10-city, it is found that our model shows 90 out of 100 cases reach the optimum and near optimum solution(within 5% error). We show the simulation of the large scale, 30-city and 50-city.

• KIEAE Journal
• /
• v.17 no.4
• /
• pp.83-88
• /
• 2017

Purpose: The objective of this study is to develop a predictive model for calculating the amount of cooling load for the different setback temperatures during the setback period. An artificial neural network (ANN) is applied as a predictive model. The predictive model is designed to be employed in the control algorithm, in which the amount of cooling load for the different setback temperature is compared and works as a determinant for finding the most energy-efficient optimal setback temperature. Method: Three major steps were conducted for proposing the ANN-based predictive model - i) initial model development, ii) model optimization, and iii) performance evaluation. Result:The proposed model proved its prediction accuracy with the lower coefficient of variation of the root mean square errors (CVRMSEs) of the simulated results (Mi) and the predicted results (Si) under generally accepted levels. In conclusion, the ANN model presented its applicability to the thermal control algorithm for setting up the most energy-efficient setback temperature.

• Wind and Structures
• /
• v.18 no.3
• /
• pp.253-265
• /
• 2014

The platform and floating structure of spar type offshore wind turbine systems should be designed in order for the 6-DOF motions to be minimized, considering diverse loading environments such as the ocean wave, wind, and current conditions. The objective of this study is to optimally design the platform and substructure of a 3MW spar type wind turbine system with the maximum postural stability in 6-DOF motions as well as the minimum material cost. Therefore, design variables of the platform and substructure were first determined and then optimized by a hydrodynamic analysis. For the hydrodynamic analysis, the body weight of the system was considered, and the ocean wave conditions were quantified to the wave forces using the Morison's equation. Moreover, the minimal number of computation analysis models was generated by the Design of Experiments (DOE), and the design variables of the platform and substructure were finally optimized by using a genetic algorithm with a neural network approximation.

• 한국전산유체공학회:학술대회논문집
• /
• 2011.05a
• /
• pp.286-290
• /
• 2011

This paper introduces the optimum macro-siting of a potential site for an offshore wind farm around Jeju Island using the RDAPS sea wind model. The statistical model was developed by analyzing the sea wind data from RDAPS model, and the meso-scale digital wind map was prepared. To develop the high resolution spatial calibration model, Artificial Neural Network(ANN) models were used to construct the wind and bathymetric maps. Accuracy and consistency of wind/bathymetric spatial calibration models were obtained using analysis of variance. The optimization problem was defined to maximize the energy density satisfying the criteria of maximum water depth and maximum distance from the coastline. The candidate site was selected through Genetic Algorithm(GA). From the results, it is possible to predict roughly a candidate site location for the installation of the offshore wind jam, and to evaluate the wind resources of the proposed site.

• 제어로봇시스템학회:학술대회논문집
• /
• 2000.10a
• /
• pp.20-20
• /
• 2000

Back-propagation neural network (BPNN) is the most prevalently used paradigm in modeling semiconductor manufacturing processes, which as a neuron activation function typically employs a bipolar or unipolar sigmoid function in either hidden and output layers. In this study, applicability of another linear function as a neuron activation function is investigated. The linear function was operated in combination with other sigmoid functions. Comparison revealed that a particular combination, the bipolar sigmoid function in hidden layer and the linear function in output layer, is found to be the best combination that yields the highest prediction accuracy. For BPNN with this combination, predictive performance once again optimized by incrementally adjusting the gradients respective to each function. A total of 121 combinations of gradients were examined and out of them one optimal set was determined. Predictive performance of the corresponding model were compared to non-optimized, revealing that optimized models are more accurate over non-optimized counterparts by an improvement of more than 30%. This demonstrates that the proposed gradient-optimized teaming for BPNN with a linear function in output layer is an effective means to construct plasma models. The plasma modeled is a hemispherical inductively coupled plasma, which was characterized by a 24 full factorial design. To validate models, another eight experiments were conducted. process variables that were varied in the design include source polver, pressure, position of chuck holder and chroline flow rate. Plasma attributes measured using Langmuir probe are electron density, electron temperature, and plasma potential.