• KIISE Transactions on Computing Practices
• /
• v.23 no.4
• /
• pp.268-273
• /
• 2017

Recently, deep learning encoder-based approach has been actively applied in the field of sentiment classification. However, Long Short-Term Memory network deep learning encoder, the commonly used architecture, lacks the quality of vector representation when the length of the documents is prolonged. In this study, for effective classification of the sentiment documents, we suggest the use of attention method-based deep learning encoder that generates document vector representation by weighted sum of the outputs of Long Short-Term Memory network based on importance. In addition, we propose methods to modify the attention method-based deep learning encoder to suit the sentiment classification field, which consist of a part that is to applied to window attention method and an attention weight adjustment part. In the window attention method part, the weights are obtained in the window units to effectively recognize feeling features that consist of more than one word. In the attention weight adjustment part, the learned weights are smoothened. Experimental results revealed that the performance of the proposed method outperformed Long Short-Term Memory network encoder, showing 89.67% in accuracy criteria.

• Journal of Nuclear Fuel Cycle and Waste Technology(JNFCWT)
• /
• v.8 no.4
• /
• pp.279-291
• /
• 2010

To characterize geological features in study area for high-level radioactive waste disposal research, KAERI (Korea Atomic Energy Research Institute) has been performing the several geological investigations such as geophysical surveys and borehole drilling since 1997. Especially, the KURT (KAERI Underground Research Tunnel) constructed to understand the deep geological environments in 2006. Recently, the deep borehole of 500 m depths was drilled to confirm and validate the geological model at the left research module of the KURT. The objective of this research was to identify the geological structures around KURT using the data obtained from the deep borehole investigation. To achieve the purpose, several geological investigations such as geophysical and borehole fracture surveys were carried out simultaneously. As a result, 7 fracture zones were identified in deep borehole located in the KURT. As one of important parts of site characterization on KURT area, the results will be used to revise the geological model of the study area.

• International Journal of Fluid Machinery and Systems
• /
• v.6 no.2
• /
• pp.56-74
• /
• 2013

Stall stagnations in the system of axial-flow compressors and ducts occur in transition from deep surge conditions to decayed or converged stall conditions. The present study is concerned with the boundaries between the deep surges and the stagnation stalls on the basis of analytical results by a code on surge transients analysis and simulation. The fundamental acoustical-geometrical stagnation boundaries were made clear from examinations of the results on a variety of duct configurations coupled with a nine-stage compressor and a single stage fan. The boundary was found to be formed by three parts, i.e., B- and A-boundaries, and an intermediate zone. The B-boundary occurs for the suction-duct having a length of about a quarter of the wave-length of the first resonance in the case of very short and fat plenum-type delivery duct. On the other hand, the A-boundary occurs for the long and narrow duct-type delivery flow-path having a length about a fifth of the wavelength and relatively small sectional area in the case of short and narrow suction ducts. In addition to this, the reduced surge-cycle frequencies with respect to the duct lengths are observed to have respective limiting values at the stagnation boundaries. The reduced frequency for the B-boundary is related with a limiting value of the Greitzer's B parameter. The tendency and the characteristic features of the related flow behaviors in the neighborhood of the boundaries were also made clearer.

• Publications of The Korean Astronomical Society
• /
• v.27 no.4
• /
• pp.367-373
• /
• 2012

The first all-sky mid-/far-infrared survey by IRAS in the 1980s, has been followed by only two more, by AKARI, from 2006, and WISE in 2010. I discuss some features of the WISE survey, and highlight some key results from early extragalactic observations that have been made by the science team during the operation of the telescope, and the post-operation proprietary period during which the public release data products were being generated. The efficient survey strategy and very high-data rate from WISE produced a catalogue of 530 million objects that was released to the public in March 2012. The WISE survey strategy naturally provided the deepest coverage at the ecliptic poles, where matched comparison fields were obtained using Spitzer, and where AKARI also observed deep fields. I describe some of the follow-up work that has been carried out based on the WISE survey, and the prospects for enhancing the WISE data by combining the AKARI survey results are also discussed. While the all-sky AKARI survey is less deep than the WISE catalogue, and is still being worked on by the AKARI science team, it includes a larger number of bands, extends to longer wavelengths, and in particular has very complementary band passes to WISE in the mid-infrared waveband, which will provide enhanced spectral information for relatively bright targets.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.14 no.9
• /
• pp.3712-3729
• /
• 2020

Smoke detection is helpful for early fire detection. With its large coverage area and low cost, vision-based smoke detection technology is the main research direction of outdoor smoke detection. We propose a two-stage smoke detection method combined with block Deep Normalization and Convolutional Neural Network (DNCNN) and visual change image. In the first stage, each suspected smoke region is detected from each frame of the images by using block DNCNN. According to the physical characteristics of smoke diffusion, a concept of visual change image is put forward in this paper, which is constructed by the video motion change state of the suspected smoke regions, and can describe the physical diffusion characteristics of smoke in the time and space domains. In the second stage, the Support Vector Machine (SVM) classifier is used to classify the Histogram of Oriented Gradients (HOG) features of visual change images of the suspected smoke regions, in this way to reduce the false alarm caused by the smoke-like objects such as cloud and fog. Simulation experiments are carried out on two public datasets of smoke. Results show that the accuracy and recall rate of smoke detection are high, and the false alarm rate is much lower than that of other comparison methods.

• 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.

• Korean Journal of Head & Neck Oncology
• /
• v.28 no.2
• /
• pp.125-128
• /
• 2012

Salivary duct carcinoma(SDC) is a highly malignant tumor of the salivary gland. The tumor is clinically characterized by a rapid onset and progression, the neoplasm is often associated with pain and facial paralysis. The nodal recurrence rate is high, and distant metastasis is common. SDC resembles high-grade breast ductal carcinoma. Curative surgical resection and postoperative radiation were the mainstay of the treatment. If facial paralysis is present, a radical parotidectomy is mandatory. Regardless of the primary location of SDC, ipsilateral functional neck dissection is indicated, because regional lymphatic spread has to be expected in the majority of patients already at time of diagnosis. If there is minor gland involvement, a bilateral neck dissection should be performed, because lymphatic drainage may occur to the contralateral side. The survival of SDC patient is poor, with most dying within three years. We experienced a unique case of SDC in parotid deep lobe. We report the clinicopathologic features of this tumor with a review of literature.

• Journal of Microbiology and Biotechnology
• /
• v.23 no.5
• /
• pp.602-613
• /
• 2013

This paper aims at an investigation of the features of bacterial communities in surface sediments of the South China Sea (SCS). In particular, biogeographical distribution patterns and the phylogenetic diversity of bacteria found in sediments collected from a coral reef platform, a continental slope, and a deep-sea basin were determined. Bacterial diversity was measured by an observation of 16S rRNA genes, and 18 phylogenetic groups were identified in the bacterial clone library. Planctomycetes, Deltaproteobacteria, candidate division OP11, and Alphaproteobacteria made up the majority of the bacteria in the samples, with their mean bacterial clones being 16%, 15%, 12%, and 9%, respectively. By comparison, the bacterial communities found in the SCS surface sediments were significantly different from other previously observed deep-sea bacterial communities. This research also emphasizes the fact that geographical factors have an impact on the biogeographical distribution patterns of bacterial communities. For instance, canonical correspondence analyses illustrated that the percentage of sand weight and water depth are important factors affecting the bacterial community composition. Therefore, this study highlights the importance of adequately determining the relationship between geographical factors and the distribution of bacteria in the world's seas and oceans.

• Journal of Korea Multimedia Society
• /
• v.20 no.6
• /
• pp.893-910
• /
• 2017

In this paper, we propose a simple online multiple object (human) tracking method, LKDeep (Lucas-Kanade feature and Detection based Simple Online Multiple Object Tracker), which can run in fast online enough on CPU core only with acceptable tracking performance for embedded surveillance purpose. The proposed LKDeep is a pragmatic hybrid approach which tracks multiple objects (humans) mainly based on LK features but is compensated by detection on periodic times or on necessity times. Compared to other state-of-the-art multiple object tracking methods based on 'Tracking-By-Detection (TBD)' approach, the proposed LKDeep is faster since it does not have to detect object on every frame and it utilizes simple association rule, but it shows a good object tracking performance. Through experiments in comparison with other multiple object tracking (MOT) methods using the public DPM detector among online state-of-the-art MOT methods reported in MOT challenge [1], it is shown that the proposed simple online MOT method, LKDeep runs faster but with good tracking performance for surveillance purpose. It is further observed through single object tracking (SOT) visual tracker benchmark experiment [2] that LKDeep with an optimized deep learning detector can run in online fast with comparable tracking performance to other state-of-the-art SOT methods.

• International Journal of Internet, Broadcasting and Communication
• /
• v.9 no.4
• /
• pp.1-7
• /
• 2017

Clipart is artificial visual contents that are created using various tools such as Illustrator to highlight some information. Here, the style of the clipart plays a critical role in determining how it looks. However, previous studies on clipart are focused only on the object recognition [16], segmentation, and retrieval of clipart images using hand-craft image features. Recently, some clipart classification researches based on the style similarity using CNN have been proposed, however, they have used different CNN-models and experimented with different benchmark dataset so that it is very hard to compare their performances. This paper presents an experimental analysis of the clipart classification based on the style similarity with two well-known CNN-models (Inception Resnet V2 [13] and VGG-16 [14] and transfers learning with the same benchmark dataset (Microsoft Style Dataset 3.6K). From this experiment, we find out that the accuracy of Inception Resnet V2 is better than VGG for clipart style classification because of its deep nature and convolution map with various sizes in parallel. We also find out that the end-to-end training can improve the accuracy more than 20% in both CNN models.