• Journal of the Korean Society for Precision Engineering
• /
• v.15 no.11
• /
• pp.152-165
• /
• 1998

This study proposes the analysis and evaluation method of time series ultrasonic signal using the chaoral post-process system for precision rate enhancement of ultrasonic pattern recognition. Chaos features extracted from time series data for analysis quantitatively weld defects For this purpose, feature extraction objectives in this study are fractal dimension, Lyapunov exponent, shape of strange attrator. Trajectory changes in the strange attractor indicated that even same type of defects carried substantial difference in chaoticity resulting from distance shifts such as nearby 0.5, 1.0 skip distance. Such difference in chaoticity enables the evaluation of unique features of defects in the weld zone. In quantitative chaos fenture extraction, feature values of 0.835 and 0.823 in the case of slag inclusion and 0.609 and 0.573 in the case of crack were suggested on the basis of fractal dimension and Lyapunov exponent. Proposed chaoral post-process system in this study can enhances precision rate of ultrasonic pattern recognition results from defect signals of weld zone, such as slag inclusion and crack.

• ETRI Journal
• /
• v.41 no.2
• /
• pp.235-241
• /
• 2019

In this paper, we propose a rank-weighted reconstruction feature to improve the robustness of a feed-forward deep neural network (FFDNN)-based acoustic model. In the FFDNN-based acoustic model, an input feature is constructed by vectorizing a submatrix that is created by slicing the feature vectors of frames within a context window. In this type of feature construction, the appropriate context window size is important because it determines the amount of trivial or discriminative information, such as redundancy, or temporal context of the input features. However, we ascertained whether a single parameter is sufficiently able to control the quantity of information. Therefore, we investigated the input feature construction from the perspectives of rank and nullity, and proposed a rank-weighted reconstruction feature herein, that allows for the retention of speech information components and the reduction in trivial components. The proposed method was evaluated in the TIMIT phone recognition and Wall Street Journal (WSJ) domains. The proposed method reduced the phone error rate of the TIMIT domain from 18.4% to 18.0%, and the word error rate of the WSJ domain from 4.70% to 4.43%.

• International journal of advanced smart convergence
• /
• v.9 no.4
• /
• pp.184-191
• /
• 2020

With the emergence of the 4th Industrial Revolution, core technologies that will lead the 4th Industrial Revolution such as AI (artificial intelligence), big data, and Internet of Things (IOT) are also at the center of the topic of the general public. In particular, there is a growing trend of attempts to present future visions by discovering new models by using them for big data analysis based on data collected in a specific field, and inferring and predicting new values with the models. In order to obtain the reliability and sophistication of statistics as a result of big data analysis, it is necessary to analyze the meaning of each variable, the correlation between the variables, and multicollinearity. If the data is classified differently from the hypothesis test from the beginning, even if the analysis is performed well, unreliable results will be obtained. In other words, prior to big data analysis, it is necessary to ensure that data is well classified according to the purpose of analysis. Therefore, in this study, data is classified using a decision tree technique and a random forest technique among classification analysis, which is a machine learning technique that implements AI technology. And by evaluating the degree of classification of the data, we try to find a way to improve the classification and analysis rate of the data.

• Journal of Korea Multimedia Society
• /
• v.24 no.12
• /
• pp.1641-1652
• /
• 2021

To analyze the crime scene, the role of digital evidence such as CCTV and black box is very important. Such digital evidence is often damaged due to device defects or intentional deletion. In this case, the deleted video can be restored by well-known techniques like the frame-based recovery method. Especially, the data such as the video can be generally fragmented and saved in the case of the memory used almost fully. If the fragmented video were recovered in units of images, the sequence of the recovered images may not be continuous. In this paper, we proposed a new video restoration method to match the sequence of recovered images. First, the images are recovered through a frame-based recovery technique. Then, after analyzing the time information marked on the images, the time information was extracted and recognized via optical character recognition (OCR). Finally, the recovered images are rearranged based on the time information obtained by OCR. For performance evaluation, we evaluate the recovery rate of our proposed video restoration method. As a result, it was shown that the recovery rate for the fragmented video was recovered from a minimum of about 47% to a maximum of 98%.

• The Journal of the Acoustical Society of Korea
• /
• v.41 no.3
• /
• pp.335-341
• /
• 2022

In this paper, we adopt a transformer that shows remarkable performance in natural language processing as an acoustic model of hybrid speech recognition. The transformer acoustic model uses attention structures to process sequential data and shows high performance with low computational cost. This paper proposes a method to improve the performance of transformer AM by applying each of the four algorithms of sequence discriminative training, a weighted finite-state transducer (wFST)-based learning used in the existing DNN-HMM model. In addition, compared to the Cross Entropy (CE) learning method, sequence discriminative method shows 5 % of the relative Word Error Rate (WER).

• Journal of the Korean Society of Food Culture
• /
• v.21 no.6
• /
• pp.577-588
• /
• 2006

This study was performed by questionnaire to investigate current use of native local foods of adults in Gyeongju classified by age. The subjects were consisted of 421 citizens(217 males and 204 females) living in Gyeongju. The findings are summarized as follows: 'Institute' scored high as 30.6% in the main responsible body for the succession of local foods. The most emphasized points to popularize the local foods was to 'taste'(36.4%). a point to be considered to develop tourism product of the local foods was to 'development of recipes acceptable to the people of today'(24.6%). The recognition rate score of native local foods of Gyeonngju area was 'Hwangnamppang', 'Hanjeongsik', 'Ssambap', 'Haejangguk', 'Hoe(Gampo)', 'Memilmukmuchim', 'Gyodongbeopju', and 'Yugoa' in the order. On the other hand, the recognition rate score for 'Ssukgulrei', 'Borisudan', 'Dalraikkakdugi', 'Hwanggeumju', 'Baesuk', 'Gyeojachae', 'Gungjungjeongol' was very low. The preferred and intake native local foods of Gyeongju area was 'Hanjeongsik', 'Ssambap', 'Hoe(Gampo)', 'Haejangguk', 'memilmukmuchim' in that other. On the other hand, the preference for 'Ssukgulrei' 'Borisudan', 'Hwanggeumju', 'Baesuk' and 'Dalraikkakdugi' was very low.

• Journal of the Korean Institute of Telematics and Electronics C
• /
• v.36C no.7
• /
• pp.46-55
• /
• 1999

We proposed the new method of postprocessing which not only reduces the frequency of dictionary access using morphological analysis but improve the recognition rate of character recognizer. In this paper, after estimating morphological construction of mis-recognized word using the part of speech that is analyzed, correct presumed mis-recognized morphology. The postprocessing using a morphology unit reduce candidate because of short than word and frequency of dictionary access because there is no need to morphological analysis for candidate. To select right candidate is only necessary to dictionary access. The proposed results show that reduced the frequency of dictionary access to 60% than postprocessing method using a word unit and recognition rate improved from 94% to 97%.

• Journal of Korea Game Society
• /
• v.16 no.2
• /
• pp.119-128
• /
• 2016

In computer games, most of GUI methods are keyboards, mouses and touch screens. The total time of processing the sound commands for games is the sum of input time and recognition time. In this paper, we propose a method for taking only the prefixes of the input signals for sound commands, resulting in the reduced the total processing time, instead of taking the whole input signals. In our method, command sounds are recognized using HMM(Hidden Markov Model), where separate HMM's are built for the whole input signals and their prefix signals. We experiment our proposed method with representative commands of platform games. The experiment shows that the total processing time of input command signals reduces without decreasing recognition rate significantly. The study will contribute to enhance the versatility of GUI for computer games.

• Journal of Fashion Business
• /
• v.16 no.3
• /
• pp.107-124
• /
• 2012

The purpose of this study was to classify types in sense of gender role identity and to figure out the difference among sociocultural attitude, benefits of clothing pursuit, and appearance management behaviors by sense of gender role identity types. Questionnaires were administered to 455 people in their twenties and thirties living in Daegu and Kyoungbuk area. Data were analyzed by using frequency, factor analysis, credibility, $X^2$-test, ANOVA, Duncan-test, and t-test. The findings are as follows. First, Men of androgyny group showed the highest rate of 38.5% followed by undifferentiation(22.9%), masculineness(21.0%), and feminineness(17.6%). Women of undifferentiation group showed the highest rate of 33.2% followed by feminineness (25.2%), androgyny(22.0%), and masculineness(19.6%). Second, factors of sociocultural attitude were internalization and recognition. Benefits of clothing pursuit consists of the factors such as consciousness of others, personality, and vogue pursuit. Factors of appearance management behaviors were dressing, skin management, plastic surgery management, weight management, health management, and hair management. Third, the result from the difference between sociocultural attitude and benefits of clothing pursuit by sense of gender role identity types, men of androgyny and feminineness showed highest in recognition while women of androgyny showed highest in internalization and recognition. The result of the difference in benefits of clothing pursuit by sense of gender role identity shows that both men and women of androgyny group have high tendency for being conscious of others, personality pursuit, and vogue pursuit while men of undifferentiation and masculineness have low interest in dressing in relationship with others. Fourth, the examination of the difference in appearance management behavior by sense of gender role identity types found men of androgyny group managed dressing, skin, weight, health and hair most while women dress and health most. Men of undifferentiation group managed dressing and weight least, while men of masculineness health, and men of feminineness hair least.

• Journal of Korea Multimedia Society
• /
• v.17 no.2
• /
• pp.124-131
• /
• 2014

In this paper, we propose the non-contact gesture recognition algorithm using 4- channel electrometer sensor array. ELF(Extremely Low Frequency) EMI and PLN are minimized because ambient electromagnetic noise around sensors has a significant impact on entire data in indoor environments. In this study, we transform AC-type data into DC-type data by applying a 10Hz LPF as well as a maximum buffer value extracting algorithm considering H/W sampling rate. In addition, we minimize the noise with the Kalman filter and extract 2-dimensional movement information by taking difference value between two cross-diagonal deployed sensors. We implemented the DTW gesture recognition algorithm using extracted data and the time delayed information of peak values. Our experiment results show that average correct classification rate is over 95% on five-gesture scenario.