• Journal of IKEEE
• /
• v.21 no.2
• /
• pp.136-141
• /
• 2017

In this paper, we propose an algorithm for memory optimization to apply deep learning to PC. The proposed algorithm minimizes the memory and computation processing time by reducing the amount of computation processing and data required in the conventional deep learning structure in a general PC. The algorithm proposed in this paper consists of three steps: a convolution layer configuration process using a random filter with discriminating power, a data reduction process using PCA, and a CNN structure creation using SVM. The learning process is not necessary in the convolution layer construction process using the discriminating random filter, thereby shortening the learning time of the overall deep learning. PCA reduces the amount of memory and computation throughput. The creation of the CNN structure using SVM maximizes the effect of reducing the amount of memory and computational throughput required. In order to evaluate the performance of the proposed algorithm, we experimented with Yale University's Extended Yale B face database. The results show that the algorithm proposed in this paper has a similar performance recognition rate compared with the existing CNN algorithm. And it was confirmed to be excellent. Based on the algorithm proposed in this paper, it is expected that a deep learning algorithm with many data and computation processes can be implemented in a general PC.

• The Research Journal of the Costume Culture
• /
• v.9 no.3
• /
• pp.458-484
• /
• 2001

The costume originated from the idea that it could protect human beings from ferocious beasts. The military costume was from protection against an enemy equipped with weapons. Furthermore, the makers of the costume were afraid of supernatural powers and it made them devise an armour and decorate it by the concept of superstition for the purpose of overcoming it. The function of military uniform was important for increasing the protection against any enemies while they join in battle. However in harmonious time, it represented artwork of peace, dignity, and order. Since the uniform was a king of symbol to show their position, and their role. They could feel proud as a member of a military group they belong to and they gain honor, courage, patriotic sentiment through their military costume. The purpose of this study was to investigate the originality the military costume as it changed and developed continuously from ancient Egypt, Greece and Rome to modern times. This study was performed to show the comparison and characteristic marks of the past, and the influence of them on the modern military uniform. The method used in this study was using past manuscripts of the early times, information attained from domestic and foreign military documents and previously written reports. As a result, the military costume was one of many means of expression of their natural and social environment. It can be said that the most epoche-marking and powerful changes to the social environment are revolution or war. War absolutely has influenced on the changes of the changes of costume and the military armour according to the type of war and the development of weapons. Especially using the amour, the main purpose that of protecting the body has changed into distinguishing our forces from the enemy and instilling fear into enemy, Furthermore, the class who has the power has used it in order to express their power and dignity. Therefore, armour had required two rules to not only protect the body but also discriminating between classes.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.14 no.12
• /
• pp.2761-2768
• /
• 2010

In this paper, we studied the use of the foreground and background features and SVM classifier to improve the accuracy of offline handwritten numeral recognition. The foreground features are two directional features: directional gradient feature by Kirsch operators and directional stroke feature by projection runlength, and the background feature is concavity feature which is extracted from the convex hull of the numeral, where concavity feature functions as complement to the directional features. During classification of the numeral, these three features are combined to obtain good discrimination power. The efficiency of our feature sets was tested by recognition experiments on the handwritten numeral database CENPARMI, where we used SVM with RBF kernel as a classifier. The experimental results showed that each combination of two or three features gave a better performance than a single feature. This means that each single feature works with a different discriminating power and cooperates with other features to enhance the recognition accuracy. By using the composite feature of the three features, we achieved a recognition rate of 98.90%.

• Proceedings of the KIEE Conference
• /
• 2006.07e
• /
• pp.55-56
• /
• 2006

In prevention and diagnostic system of GIS, pattern classification is focused on the detection of unnatural patterns in PD(Partial discharge) image data. Fractals have been used extensively to provide a description and to model mathematically many of the naturally occurring complex shapes, such as coastlines, mountain ranges, clouds, etc., and have also received increased attention in the field of image processing, for purposes of segmentation and recognition of regions and objects present in natural scenes. Among the numerous fractal features that could be defined and used for image data, fractal dimension and lacunarity have been found to be useful for recognition purposes Partial discharge(PD) occuring in GIS system is a very complex phenomenon, and more so are the shapes of the various 2-d patterns obtained during routine tests and measurements. It has been fairly well established that these pattern shapes and underlying defects causing PD have a 1:1 correspondence, and therefore methods to describe and qunatify these pattern shapes must be explored, before recognition systems based on them could be developed. The computed fractal features(fractal dimension and lacunarity) for standard library of PD data were analyzed and found to possess fairly reasonable pattern discriminating abilities. This new approach appears promising, and further research is essential before any long-term predictions can be made.

• Journal of Biomedical Engineering Research
• /
• v.33 no.1
• /
• pp.15-24
• /
• 2012

Classification of human thought is an emerging research field that may allow us to understand human brain functions and further develop advanced brain-computer interface (BCI) systems. In the present study, we introduce a new approach to classify various mental states from noninvasive electrophysiological recordings of human brain activity. We utilized the full spatial and spectral information contained in the electroencephalography (EEG) signals recorded while a subject is performing a specific mental task. For this, the EEG data were converted into a 2D spatiospectral pattern map, of which each element was filled with 1, 0, and -1 reflecting the degrees of event-related synchronization (ERS) and event-related desynchronization (ERD). We evaluated the similarity between a current (input) 2D pattern map and the template pattern maps (database), by taking the inner-product of pattern matrices. Then, the current 2D pattern map was assigned to a class that demonstrated the highest similarity value. For the verification of our approach, eight participants took part in the present study; their EEG data were recorded while they performed four different cognitive imagery tasks. Consistent ERS/ERD patterns were observed more frequently between trials in the same class than those in different classes, indicating that these spatiospectral pattern maps could be used to classify different mental states. The classification accuracy was evaluated for each participant from both the proposed approach and a conventional mental state classification method based on the inter-hemispheric spectral power asymmetry, using the leave-one-out cross-validation (LOOCV). An average accuracy of 68.13% (${\pm}9.64%$) was attained for the proposed method; whereas an average accuracy of 57% (${\pm}5.68%$) was attained for the conventional method (significance was assessed by the one-tail paired $t$-test, $p$ < 0.01), showing that the proposed simple classification approach might be one of the promising methods in discriminating various mental states.

• The Korean Journal of Cytopathology
• /
• v.7 no.2
• /
• pp.169-176
• /
• 1996

This study was peformed in order to evaluate the usefulness of the crush cytologic features and differential diagnosis between meningiomas and schwannomas in the central nervous system. Deeply seated and unusually located meningiomas and schwannomas with equivocal or erroneous frozen section diagnosis can be correctly diagnosed cytologically in crush preparations. Twenty-four meningiomas and nine schwannomas were studied by frozen section and crush preparation technique. These tumors displayed distinctive cytologic features. in meningiomas, the tumor tissue fragments were easy to crush, and the tumor cells were arranged in small clusters, flat sheets, papilla-like, whorling pattern or singly. Individual tumor cells displayed round or oval nuclei with finely granular chromatin pattern and inconspicuous small nucleoli. Occasionally psammoma bodies, nuclear pseudoinclusion or nuclear grooves were found. In schwannomas, tissue fragments were hard in consistency and difficult to crush. The crushed tissue presented as thick, irregular fragments with shard borders. The cells showed ill-defined cytoplasm and round, oval, cigar-shaped or curved nuclei. It is important to emphasize that the smear pattern under low-power view and cytologic features are helpful in discriminating between these two tumors.

• The Korean Journal of Malacology
• /
• v.26 no.4
• /
• pp.275-283
• /
• 2010

Many freshwater snail taxa are difficult to identify using morphological traits due to phenotypic plasticity. However, using of molecular DNA marker in combination with morphological traits can provide a reliable means for discriminating among freshwater snail taxa including cryptic species. To discriminate among Korean freshwater snail taxa and resolve their systematic relationships, wesequenced a fragment of mtDNA cytochrome oxidase I (COI) gene from 82 specimens collected from ten different sites distributed along the Korean peninsula. We identified more than seven freshwater snail taxa including cryptic species in Korea. Whereas traditional shell morphology of freshwater snails offers only weak discriminatory power for recognizing 'good' taxa, DNA sequence data provided positive and reliable identification. In addition, a major Semisulcospira clade was clearly separated from the remaining lineages observed including cryptic species. However, a phylogenetic tree inferred from the COI gene data did not fully resolve systematic relationships among pleurocerid taxa in Korea. Establishing more robust shell characteristics for identifying taxa unambiguously and hence improving traditional key shell morphology characters for freshwater snail species is an urgent requirement and will require more rigorous examination of all nominal taxa. While molecular data generated here will be useful for species identification and for describing the systematic relationships among Korean freshwater snails, further analysis will be required.

• Journal of the Korean Society for information Management
• /
• v.34 no.3
• /
• pp.209-228
• /
• 2017

This study suggests a new Hirsch-type composite index, 'transposed g-index' with a different viewpoint on h-index and g-index. From this new point of view, the axes of the graph describing the h-index and g-index are transposed so that the horizontal axis corresponds to the citation frequency threshold and the vertical axis corresponds to the number of documents. Based on this transposed graph, a new indicator transposed g-index is suggested and applied to library and information science researchers' outcomes in Korean Citation Index database. The results show that this new index has more discriminating power than h-index and g-index, and is more sensitive to differences in quantitative aspects than quality of research. It is expected that the transposed g-index will be helpful for the multifaceted evaluation of the research outcome because it has differentiating characteristics that distinguish consistent researchers who continue to study from those who do not.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.33 no.11A
• /
• pp.1087-1094
• /
• 2008

In this paper, we propose the acquisition method of the Cyclic-Prefix CDMA using Good-Autocorrelation-Sequence (GAS-CP-CDMA) system. The sequences for GAS-CP-CDMA are m-sequence and constant amplitude zero autocorrelation code (CAZAC) with good autocorrelation property. The GAS-CP-CDMA system is proposed for interference cancellation in multi-user and -path environment in co-cell. However, the reliable initial acquisition in multi-path fading channel and multi-user system environment is crucial. In this paper, non-coherent power detector and threshold value are used for discriminating signal acquisition. The sequence used for computer simulation is CAZAC. The simulation results show proper threshold value depends on the number of users. Through the computer simulation, we draw setting up method of the optimum threshold value for the GAS-CP-CDMA system.

• Applied Microscopy
• /
• v.38 no.3
• /
• pp.185-193
• /
• 2008

An experimental comparison of the detection properties between imaging plate and film for recording the electron diffraction pattern was carried out on a radiation-sensitive material, an aluminum trihydroxide(gibbsite, ${\gamma}-Al(OH)_3$), through the electron beam irradiation. Because the imaging plate has a wide dynamic range sufficient for recording extremely low- and high-electron intensities, the range of spatial frequency for the diffraction pattern acquired by the imaging plate was extended to two times larger than the range by the film, especially at a low electron dose condition(${\leq}0.1\;e^-/{\mu}m^2$). It is also demonstrated that the imaging plate showed better resolving power for discriminating fine intensity levels even in saturated transmitted beam. Hence, in the respect of investigating the structures of radiation-sensitive materials and cryo-biological specimens, our experimental demonstrations suggest that the imaging plate technique may be a good choice for those studies, which have to use an extremely low electron intensity for recording.