• Journal of KIISE
• /
• v.44 no.3
• /
• pp.306-313
• /
• 2017

Named entity recognition (NER) seeks to locate and classify named entities in text into pre-defined categories such as names of persons, organizations, locations, expressions of times, etc. Recently, many state-of-the-art NER systems have been implemented with bidirectional LSTM CRFs. Deep learning models based on long short-term memory (LSTM) generally depend on word representations as input. In this paper, we propose an approach to expand word representation by using pre-trained word embedding, part of speech (POS) tag embedding, syllable embedding and named entity dictionary feature vectors. Our experiments show that the proposed approach creates useful word representations as an input of bidirectional LSTM CRFs. Our final presentation shows its efficacy to be 8.05%p higher than baseline NERs with only the pre-trained word embedding vector.

• Journal of Digital Convergence
• /
• v.13 no.12
• /
• pp.441-451
• /
• 2015

The research objective of this study to provide a basis for university basketball players development university basketball players have recovered and fair confidence in the result of the determination of analyzes feel the referee judges recognized research players and referees and by the referee judged that consistency there is. The referee judged according to gender awareness and the impact analysis of college basketball players impartiality, promptness, consistency, reliability p <.05 level There was a statistically significant difference. The referee judged recognition and impact of the exhibition was a statistically significant difference in the p <.05 level of agility. The referee judged recognition and influence of the position was no statistically significant difference in the p <.05 level of agility. impact of the judgment determining recognition fairness, promptness, there was a statistically significant difference in the p <.001 level of reliability. The referee judged of the psychological state is assigned prejudice, there was a statistically significant difference in the p <.001 level.

• Fire Science and Engineering
• /
• v.27 no.6
• /
• pp.77-82
• /
• 2013

This research suggested smart fire recognition system which trances firing location with CCD camera with wired/wire-less TCP/IP function and Pan/Tilt function, delivers information in real time to android system installed by smart mobile communication system and controls fire and disaster remotely. To embody suggested method, firstly, algorithm which applies hue saturation intensity (HSI) Transform for input video, eliminates surrounding lightness and unnecessary videos and segmentalized only firing videos was suggested. Secondly, Pan/Tilt function traces accurate location of firing for proper control of firing. Thirdly, android communication system installed by mobile function confirms firing state and controls it. To confirm the suggested method, 10 firing videos were input and experiment was conducted. As the result, all of 10 videos segmentalized firing sector and traced all of firing locations.

• KIISE Transactions on Computing Practices
• /
• v.23 no.7
• /
• pp.417-423
• /
• 2017

Recently, many embedded devices that have the computing capability required for deep learning have become available; hence, many new applications using these devices are emerging. However, these embedded devices have an architecture different from that of PCs and high-performance servers. In this paper, we propose a method that improves the performance of deep-learning framework by considering the architecture of an embedded device that shares memory between the CPU and the GPU. The proposed method is implemented in Caffe, an open-source deep-learning framework, and is evaluated on an NVIDIA Jetson TK1 embedded device. In the experiment, we investigate the image recognition performance of several state-of-the-art deep-learning networks, including AlexNet, VGGNet, and GoogLeNet. Our results show that the proposed method can achieve significant performance gain. For instance, in AlexNet, we could reduce image recognition latency by about 33% and energy consumption by about 50%.

• Journal of the Korean Institute of Illuminating and Electrical Installation Engineers
• /
• v.22 no.4
• /
• pp.26-34
• /
• 2008

This study focus on segmentation process that classifies road surfaces into 5 different categories, dry, wet water, icy, and snowy surfaces by analyzing asphalt-paved road images taken in daylight. By using the polarization coefficients, the proportions of horizontally polarized components to vertically polarized components, regions with over 1.3 polarization coefficients are classified as wet surfaces. Except for wet surfaces, the decision process a lies time-frequency analysis to other parts by using the third order wavelet packet transform. In addition, by using the average frequency characteristics of dry and icy surfaces from image templates, decide which is closer to a test image, and finally identify dry and icy surfaces. It is confirmed that the reposed estimation and segmentation of recognition on various images. This can be interpreted as an indication that image-only mad surface condition supervision is probable.

• Journal of Korean society of Dental Hygiene
• /
• v.6 no.1
• /
• pp.63-77
• /
• 2006

The purpose of this study was to prepare a student and education guideline by examining the college life satisfaction of students studying dental hygiene. Subjects were students going to dental hygiene-related department of three colleges in the Gwangju and Jeonnam areas. College life and life satisfaction was examined at general characteristics. The following results were obtained. 1. Subjects included 35.7% of 1st grade, 33.4% of 2nd grade, and 30.9% of 3rd grade. In the health state, average accounted for the highest percent(76.9%); in the student's associates, a little for the highest percent(76.8%); and in the participation in clubs, no participation for the highest percent(77.1%). 2. For motives to select the dental hygiene, good employment after graduation accounted for the highest percent(61.0%). The highest desired employer was a dental clinic(33.5%), and the highest desired duration was a whole life(61.0%). 3. According to the grade, satisfaction was significantly different in social recognition, major in education, professor, facilities, professor-student, career, and college development(pE0.05). 4. According to the school record, satisfaction was significantly different in social recognition, major in education, professor, professor-student, career, college development, and general life(pE0.01). 5. According to the student's associates, satisfaction was significantly different in social recognition, major in education, professor-student, and general life(pE0.01). 6. According to the participation in clubs, satisfaction was significantly different in professor, facilities, college development, and religious life(pE0.05). In conclusion, as students were in the lower grade, made good school record, had happier associates, and participated in clubs more actively, their satisfaction at college life became higher. Accordingly, professors and school authorities should strive to examine students' actual problem and discuss and resolve them in order to help their effective school life.

• The Journal of the Acoustical Society of Korea
• /
• v.23 no.2
• /
• pp.90-102
• /
• 2004

According to the Mutual Recognition Arrangement (MRA), the calibration and/or test laboratories should satisfy the management and technical requirements ISO 17025 or equivalent. Chapter 5, Section 5.10.4 of the technical requirement of ISO 17025 suggests the required informations for calibration certificates, one of them is to state the uncertainty of measurement. The uncertainties of measurement in reciprocity calibration of standard laboratory microphone were calculated. The expanded uncertainties for 1 and 1/2 inch microphones were 0.03 dB in the middle frequency range and they increased up to 0.10 dB and 0.11 dB at 20 Hz, 0.07 dB and 0.08 dB at high frequency, respectively.

• Journal of Korea Multimedia Society
• /
• v.10 no.8
• /
• pp.981-996
• /
• 2007

In this paper, we describe a method that can recognize gestures by obtaining motion features information with principal factor analysis from sequential gesture images. In the algorithm, firstly, a two dimensional silhouette region including human gesture is segmented and then geometric features are extracted from it. Here, global features information which is selected as some meaningful key feature effectively expressing gestures with principal factor analysis is used. Obtained motion history information representing time variation of gestures from extracted feature construct one gesture subspace. Finally, projected model feature value into the gesture space is transformed as specific state symbols by grouping algorithm to be use as input symbols of HMM and input gesture is recognized as one of the model gesture with high probability. Proposed method has achieved higher recognition rate than others using only shape information of human body as in an appearance-based method or extracting features intuitively from complicated gestures, because this algorithm constructs gesture models with feature factors that have high contribution rate using principal factor analysis.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.21 no.4
• /
• pp.737-743
• /
• 2017

The intelligent life-log system proposed in this paper is intended to identify and record a myriad of everyday life information as to the occurrence of various events based on when, where, with whom, what and how, that is, a wide variety of contextual information involving person, scene, ages, emotion, relation, state, location, moving route, etc. with a unique tag on each piece of such information and to allow users to get a quick and easy access to such information. Context awareness generates and classifies information on a tag unit basis using the auto-tagging technology and biometrics recognition technology and builds a situation information database. In this paper, we developed an active modeling method and an application that recognizes expressionless and smile expressions using lip lines to automatically record emotion information.

• Journal of the Institute of Electronics and Information Engineers
• /
• v.51 no.6
• /
• pp.89-101
• /
• 2014

In this paper, we proposed a new local micro pattern, Signed Local Directional Pattern(SLDP). SLDP uses information of edges to represent the face's texture. This can produce a more discriminating and efficient code than other state-of-the-art methods. Each micro pattern of SLDP is encoded by sign and its major directions in which maximum edge responses exist-which allows it to distinguish among similar edge patterns that have different intensity transitions. In this paper, we divide the face image into several regions, each of which is used to calculate the distributions of the SLDP codes. Each distribution represents features of the region and these features are concatenated into a feature vector. We carried out facial expression recognition with feature vectors and SVM(Support Vector Machine) on Cohn-Kanade and JAFFE databases. SLDP shows better classification accuracy than other existing methods.