• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.18 no.10
• /
• pp.75-80
• /
• 2017

In this paper, we describe the effect of the regularization method and the network with identity mapping on the performance of the activation functions in deep convolutional neural networks. The activation functions act as nonlinear transformation. In early convolutional neural networks, a sigmoid function was used. To overcome the problem of the existing activation functions such as gradient vanishing, various activation functions were developed such as ReLU, Leaky ReLU, parametric ReLU, and ELU. To solve the overfitting problem, regularization methods such as dropout and batch normalization were developed on the sidelines of the activation functions. Additionally, data augmentation is usually applied to deep learning to avoid overfitting. The activation functions mentioned above have different characteristics, but the new regularization method and the network with identity mapping were validated only using ReLU. Therefore, we have experimentally shown the effect of the regularization method and the network with identity mapping on the performance of the activation functions. Through this analysis, we have presented the tendency of the performance of activation functions according to regularization and identity mapping. These results will reduce the number of training trials to find the best activation function.

• Journal of Institute of Control, Robotics and Systems
• /
• v.21 no.12
• /
• pp.1142-1151
• /
• 2015

In this paper, High performance torque control based on the flux mapping of 48V Wound Rotor Synchronous Motor has been studied to improve torque control. Flux map considering MTPA (Maximum Torque Per Ampere), MFPT (Minimum Flux Per Torque), Maximum efficiency point at the same torque command and flux command for each field current was produced. Current map using flux mapping of Each field current was applied to the MTPA, MFPT. Generating a current vector locus was to determine the characteristics of the operation region. Through the Matlab/Simulink simulation, difference between speed-torque map and flux map was represented. The suggested flux map was tested actual experiments on a dynamometer.

• Journal of Information Processing Systems
• /
• v.13 no.3
• /
• pp.435-447
• /
• 2017

Associative and bidirectional associative memories are examples of associative structures studied intensively in the literature. The underlying idea is to realize associative mapping so that the recall processes (one-directional and bidirectional ones) are realized with minimal recall errors. Associative and fuzzy associative memories have been studied in numerous areas yielding efficient applications for image recall and enhancements and fuzzy controllers, which can be regarded as one-directional associative memories. In this study, we revisit and augment the concept of associative memories by offering some new design insights where the corresponding mappings are realized on the basis of a related collection of landmarks (prototypes) over which an associative mapping becomes spanned. In light of the bidirectional character of mappings, we have developed an augmentation of the existing fuzzy clustering (fuzzy c-means, FCM) in the form of a so-called collaborative fuzzy clustering. Here, an interaction in the formation of prototypes is optimized so that the bidirectional recall errors can be minimized. Furthermore, we generalized the mapping into its granular version in which numeric prototypes that are formed through the clustering process are made granular so that the quality of the recall can be quantified. We propose several scenarios in which the allocation of information granularity is aimed at the optimization of the characteristics of recalled results (information granules) that are quantified in terms of coverage and specificity. We also introduce various architectural augmentations of the associative structures.

• The Journal of Korean Institute of Electromagnetic Engineering and Science
• /
• v.19 no.11
• /
• pp.1248-1253
• /
• 2008

The MCS(Micro-Coplanar Strip) line with ground has been analyzed. The conformal mapping method is used to calculate the quasi-static effective dielectric constant and characteristic impedance of this MCS line. The computed results of the present work are found to be in good agreement when compared with the results obtained using commercial S/W, IE3D. And in this paper, the stepped-impedance low pass filter is designed and fabricated with MCS lines for improving the frequency responses. The LPF proposed structure has been also designed and implemented to have the sharp attenuation characteristics in stop band. The agreement between simulation and measurement results verify the implemented LPF.

• Journal of KIISE:Software and Applications
• /
• v.30 no.1_2
• /
• pp.1-18
• /
• 2003

Precise and systematic mapping techniques are required for mapping object-oriented artifacts into a platform-specific design. An effective and systematic mapping approach for an adequate platform or programming language in needed, because the characteristics of an initial design are independent from an implementation language and a platform. In this paper, we propose systematic and concrete methods, guidelines, and design patterns that can be used to design business operations at EJB (Enterprise JavaBeans) source code level. We show how various EJB mechanism can be utilized in designing business operations for beans. We believe these proposed methods can yield high-performance EJB applications that can also be well maintainable.

• Journal of the Korean Society of Combustion
• /
• v.8 no.3
• /
• pp.15-23
• /
• 2003

Partial quenching structure of turbulent diffusion flames in a turbulent mixing layer is investigated by the method of flame hole dynamics in order to develop a prediction model for turbulent flame lift off. The essence of flame hole dynamics is derivation of the random walk mapping, from the flame-edge theory, which governs expansion or contraction of flame holes initially created by local quenching events. The numerical simulation for flame hole dynamics is carried out in two stages. First, a direct numerical simulation is performed for constant-density fuel-air channel mixing layer to obtain the turbulent flow and mixing fields, from which a time series of two dimensional scalar dissipation rate array is extracted at a fixed virtual flame surface horizontally extending from the end of split plate to the downstream. Then, the Lagrangian simulation of the flame hole random walk mapping projected to the scalar dissipation rate array yields temporally evolving turbulent extinction process and its statistics on partial quenching characteristics. The statistical results exhibit that the chance of partial quenching is strongly influenced by the crossover scalar dissipation rate while almost unaffected by the iteration number of the mapping that can be regarded as a flame-edge speed.

• Journal of the Korean Society of Surveying, Geodesy, Photogrammetry and Cartography
• /
• v.34 no.6
• /
• pp.597-607
• /
• 2016

The goal of this study was to analyze landslide susceptibility using two different models and compare the results. For this purpose, a landslide inventory map was produced from a field survey, and the inventory was divided into two groups for training and validation, respectively. Sixteen landslide conditioning factors were considered. The relationships between landslide occurrence and landslide conditioning factors were analyzed using the FR (Frequency Ratio) and EBF (Evidential Belief Function) models. The LSI (Landslide Susceptibility Index) maps that were produced were validated using the ROC (Relative Operating Characteristics) curve and the SCAI (Seed Cell Area Index). The AUC (Area under the ROC Curve) values of the FR and EBF LSI maps were 80.6% and 79.5%, with prediction accuracies of 72.7% and 71.8%, respectively. Additionally, in the low and very low susceptibility zones, the FR LSI map had higher SCAI values compared to the EBF LSI map, as high as 0.47%p. These results indicate that both models were reasonably accurate, however that the FR LSI map had a slightly higher accuracy for landslide susceptibility mapping in the study area.

• Journal of the Korean Chemical Society
• /
• v.61 no.1
• /
• pp.12-15
• /
• 2017

We presented a mapping the work function of the vanadium pentoxide ($V_2O_5$) nanonet structures by scanning Kelvin probe microscopy (SKPM). In this measurement, the $V_2O_5$ nanonet was self-assembled via dropping the solution of $V_2O_5$ nanowires (NWs) onto the $SiO_2$ substrate and drying the solvent, resulting in the networks of $V_2O_5$ NWs. We found that the SKPM signal as a surface potential of $V_2O_5$ nanonet is attributed to the contact potential difference (CPD) between the work functions of the metal tip and the $V_2O_5$ nanonet. We generated the histograms of the CPD signals obtained from the SKPM mapping of the $V_2O_5$ nanonet as well as the highly ordered pyrolytic graphite (HOPG) which is used as a reference for the calibration of the SKPM tip. By using the histogram peaks of the CPD signals, we successfully estimated the work function of ~5.1 eV for the $V_2O_5$ nanonet structures. This work provides a possibility of a nanometer-scale imaging of the work function of the various nanostructures and helps to understand the electrical characteristics of the future electronic devices.

• Proceedings of the Korean Institute Of Construction Engineering and Management
• /
• 2008.11a
• /
• pp.497-500
• /
• 2008

Recently, the urban regeneration project has been performed actively at home and abroad. The stake-holders in urban regeneration project are various and complicated, and has large scale during a long life period. Also they show the characteristics of a mega-project and most mixed-use form. Therefore, the urban regeneration has a lot of potential risk factors from project beginning to completion. It means they need efficient and continuous risk management in terms of performance measurement. But the current domestic construction project does not reflect risk management in view of performance measurement. This study proposes the risk management methodology by mapping risk factors with major performance indexes of the urban regeneration project.

• JSTS:Journal of Semiconductor Technology and Science
• /
• v.14 no.4
• /
• pp.391-406
• /
• 2014

The performance of General-Purpose computation on Graphics Processing Units (GPGPU) is heavily dependent on the memory access behavior. This sensitivity is due to a combination of the underlying Massively Parallel Processing (MPP) execution model present on GPUs and the lack of architectural support to handle irregular memory access patterns. Application performance can be significantly improved by applying memory-access-pattern-aware optimizations that can exploit knowledge of the characteristics of each access pattern. In this paper, we present an algorithmic methodology to semi-automatically find the best mapping of memory accesses present in serial loop nest to underlying data-parallel architectures based on a comprehensive static memory access pattern analysis. To that end we present a simple, yet powerful, mathematical model that captures all memory access pattern information present in serial data-parallel loop nests. We then show how this model is used in practice to select the most appropriate memory space for data and to search for an appropriate thread mapping and work group size from a large design space. To evaluate the effectiveness of our methodology, we report on execution speedup using selected benchmark kernels that cover a wide range of memory access patterns commonly found in GPGPU workloads. Our experimental results are reported using the industry standard heterogeneous programming language, OpenCL, targeting the NVIDIA GT200 architecture.