• Journal of the Korea Society of Computer and Information
• /
• v.28 no.10
• /
• pp.27-35
• /
• 2023

In this paper, we propose a process of increasing productivity by applying a deep learning-based defect detection and classification system to the prepreg fiber manufacturing process, which is in high demand in the field of producing composite materials. In order to apply it to toe prepreg manufacturing equipment that requires a solution due to the occurrence of a large amount of defects in various conditions, the optimal environment was first established by selecting cameras and lights necessary for defect detection and classification model production. In addition, data necessary for the production of multiple classification models were collected and labeled according to normal and defective conditions. The multi-classification model is made based on CNN and applies pre-learning models such as VGGNet, MobileNet, ResNet, etc. to compare performance and identify improvement directions with accuracy and loss graphs. Data augmentation and dropout techniques were applied to identify and improve overfitting problems as major problems. In order to evaluate the performance of the model, a performance evaluation was conducted using the confusion matrix as a performance indicator, and the performance of more than 99% was confirmed. In addition, it checks the classification results for images acquired in real time by applying them to the actual process to check whether the discrimination values are accurately derived.

• Journal of the Computational Structural Engineering Institute of Korea
• /
• v.37 no.1
• /
• pp.9-16
• /
• 2024

Using the proper orthogonal decomposition (POD) based intrusive reduced order model (ROM), the total degrees of freedom of the structural system can be significantly reduced and the critical time step satisfying the conditional stability increases in the explicit time integrations. In this study, therefore, the changes in the critical time step in the explicit time integrations are investigated using both the POD-ROM and Voronoi-cell lattice model (VCLM). The snapshot matrix is composed of the data from the structural response under the arbitrary dynamic loads such as seismic excitation, from which the POD-ROM is constructed and the predictive capability is validated. The simulated results show that the significant reduction in the computational time can be achieved using the POD-ROM with sufficiently ensuring the numerical accuracy in the seismic analyses. In addition, the validations show that the POD based intrusive ROM is compatible with the Voronoi-cell lattice based explicit dynamic analyses. In the future study, the research results will be utilized as an elemental technology for the developments of the real-time predictive models or monitoring system involving the high-fidelity simulations of structural dynamics.

• The Transactions of the Korea Information Processing Society
• /
• v.13 no.9
• /
• pp.444-452
• /
• 2024

This study proposes a deep learning architecture optimized for fire detection derived through Layer Importance Evaluation. In order to solve the problem of unnecessary complexity and operation of the existing Convolutional Neural Network (CNN)-based fire detection system, the operation of the inner layer of the model based on the weight and activation values was analyzed through the Layer Importance Evaluation technique, the layer with a high contribution to fire detection was identified, and the model was reconstructed only with the identified layer, and the performance indicators were compared and analyzed with the existing model. After learning the fire data using four transfer learning models: Xception, VGG19, ResNet, and EfficientNetB5, the Layer Importance Evaluation technique was applied to analyze the weight and activation value of each layer, and then a new model was constructed by selecting the top rank layers with the highest contribution. As a result of the study, it was confirmed that the implemented architecture maintains the same performance with parameters that are about 80% lighter than the existing model, and can contribute to increasing the efficiency of fire monitoring equipment by outputting the same performance in accuracy, loss, and confusion matrix indicators compared to conventional complex transfer learning models while having a learning speed of about 3 to 5 times faster.

• Journal of Intelligence and Information Systems
• /
• v.22 no.3
• /
• pp.113-127
• /
• 2016

Among the techniques for recommendation, collaborative filtering (CF) is commonly recognized to be the most effective for implementing recommender systems. Until now, CF has been popularly studied and adopted in both academic and real-world applications. The basic idea of CF is to create recommendation results by finding correlations between users of a recommendation system. CF system compares users based on how similar they are, and recommend products to users by using other like-minded people's results of evaluation for each product. Thus, it is very important to compute evaluation similarities among users in CF because the recommendation quality depends on it. Typical CF uses user's explicit numeric ratings of items (i.e. quantitative information) when computing the similarities among users in CF. In other words, user's numeric ratings have been a sole source of user preference information in traditional CF. However, user ratings are unable to fully reflect user's actual preferences from time to time. According to several studies, users may more actively accommodate recommendation of reliable others when purchasing goods. Thus, trust relationship can be regarded as the informative source for identifying user's preference with accuracy. Under this background, we propose a new hybrid recommender system that fuses CF and social network analysis (SNA). The proposed system adopts the recommendation algorithm that additionally reflect the result analyzed by SNA. In detail, our proposed system is based on conventional memory-based CF, but it is designed to use both user's numeric ratings and trust relationship information between users when calculating user similarities. For this, our system creates and uses not only user-item rating matrix, but also user-to-user trust network. As the methods for calculating user similarity between users, we proposed two alternatives - one is algorithm calculating the degree of similarity between users by utilizing in-degree and out-degree centrality, which are the indices representing the central location in the social network. We named these approaches as 'Trust CF - All' and 'Trust CF - Conditional'. The other alternative is the algorithm reflecting a neighbor's score higher when a target user trusts the neighbor directly or indirectly. The direct or indirect trust relationship can be identified by searching trust network of users. In this study, we call this approach 'Trust CF - Search'. To validate the applicability of the proposed system, we used experimental data provided by LibRec that crawled from the entire FilmTrust website. It consists of ratings of movies and trust relationship network indicating who to trust between users. The experimental system was implemented using Microsoft Visual Basic for Applications (VBA) and UCINET 6. To examine the effectiveness of the proposed system, we compared the performance of our proposed method with one of conventional CF system. The performances of recommender system were evaluated by using average MAE (mean absolute error). The analysis results confirmed that in case of applying without conditions the in-degree centrality index of trusted network of users(i.e. Trust CF - All), the accuracy (MAE = 0.565134) was lower than conventional CF (MAE = 0.564966). And, in case of applying the in-degree centrality index only to the users with the out-degree centrality above a certain threshold value(i.e. Trust CF - Conditional), the proposed system improved the accuracy a little (MAE = 0.564909) compared to traditional CF. However, the algorithm searching based on the trusted network of users (i.e. Trust CF - Search) was found to show the best performance (MAE = 0.564846). And the result from paired samples t-test presented that Trust CF - Search outperformed conventional CF with 10% statistical significance level. Our study sheds a light on the application of user's trust relationship network information for facilitating electronic commerce by recommending proper items to users.

• Journal of Intelligence and Information Systems
• /
• v.27 no.3
• /
• pp.175-197
• /
• 2021

Recently, with the development of deep learning technology, research on unstructured data analysis is being actively conducted, and it is showing remarkable results in various fields such as classification, summary, and generation. Among various text analysis fields, text classification is the most widely used technology in academia and industry. Text classification includes binary class classification with one label among two classes, multi-class classification with one label among several classes, and multi-label classification with multiple labels among several classes. In particular, multi-label classification requires a different training method from binary class classification and multi-class classification because of the characteristic of having multiple labels. In addition, since the number of labels to be predicted increases as the number of labels and classes increases, there is a limitation in that performance improvement is difficult due to an increase in prediction difficulty. To overcome these limitations, (i) compressing the initially given high-dimensional label space into a low-dimensional latent label space, (ii) after performing training to predict the compressed label, (iii) restoring the predicted label to the high-dimensional original label space, research on label embedding is being actively conducted. Typical label embedding techniques include Principal Label Space Transformation (PLST), Multi-Label Classification via Boolean Matrix Decomposition (MLC-BMaD), and Bayesian Multi-Label Compressed Sensing (BML-CS). However, since these techniques consider only the linear relationship between labels or compress the labels by random transformation, it is difficult to understand the non-linear relationship between labels, so there is a limitation in that it is not possible to create a latent label space sufficiently containing the information of the original label. Recently, there have been increasing attempts to improve performance by applying deep learning technology to label embedding. Label embedding using an autoencoder, a deep learning model that is effective for data compression and restoration, is representative. However, the traditional autoencoder-based label embedding has a limitation in that a large amount of information loss occurs when compressing a high-dimensional label space having a myriad of classes into a low-dimensional latent label space. This can be found in the gradient loss problem that occurs in the backpropagation process of learning. To solve this problem, skip connection was devised, and by adding the input of the layer to the output to prevent gradient loss during backpropagation, efficient learning is possible even when the layer is deep. Skip connection is mainly used for image feature extraction in convolutional neural networks, but studies using skip connection in autoencoder or label embedding process are still lacking. Therefore, in this study, we propose an autoencoder-based label embedding methodology in which skip connections are added to each of the encoder and decoder to form a low-dimensional latent label space that reflects the information of the high-dimensional label space well. In addition, the proposed methodology was applied to actual paper keywords to derive the high-dimensional keyword label space and the low-dimensional latent label space. Using this, we conducted an experiment to predict the compressed keyword vector existing in the latent label space from the paper abstract and to evaluate the multi-label classification by restoring the predicted keyword vector back to the original label space. As a result, the accuracy, precision, recall, and F1 score used as performance indicators showed far superior performance in multi-label classification based on the proposed methodology compared to traditional multi-label classification methods. This can be seen that the low-dimensional latent label space derived through the proposed methodology well reflected the information of the high-dimensional label space, which ultimately led to the improvement of the performance of the multi-label classification itself. In addition, the utility of the proposed methodology was identified by comparing the performance of the proposed methodology according to the domain characteristics and the number of dimensions of the latent label space.

• Journal of Intelligence and Information Systems
• /
• v.29 no.3
• /
• pp.1-17
• /
• 2023

With the recent rapid development of ICT(Information and Communication Technology) and the popularization of digital devices, the size of the online market continues to grow. As a result, we live in a flood of information. Thus, customers are facing information overload problems that require a lot of time and money to select products. Therefore, a personalized recommender system has become an essential methodology to address such issues. Collaborative Filtering(CF) is the most widely used recommender system. Traditional recommender systems mainly utilize quantitative data such as rating values, resulting in poor recommendation accuracy. Quantitative data cannot fully reflect the user's preference. To solve such a problem, studies that reflect qualitative data, such as review contents, are being actively conducted these days. To quantify user review contents, text mining was used in this study. The general CF consists of the following three steps: user-item matrix generation, Top-N neighborhood group search, and Top-K recommendation list generation. In this study, we propose a recommendation algorithm that applies an extended similarity measure, which utilize quantified review contents in addition to user rating values. After calculating review similarity by applying TF-IDF, Word2Vec, and Doc2Vec techniques to review content, extended similarity is created by combining user rating similarity and quantified review contents. To verify this, we used user ratings and review data from the e-commerce site Amazon's "Health and Personal Care". The proposed recommendation model using extended similarity measure showed superior performance to the traditional recommendation model using only user rating value-based similarity measure. In addition, among the various text mining techniques, the similarity obtained using the TF-IDF technique showed the best performance when used in the neighbor group search and recommendation list generation step.

• Journal of Electrical Engineering and Technology
• /
• v.5 no.2
• /
• pp.342-351
• /
• 2010

The SiC-$ZrB_2$ composites were fabricated by combining 30, 35, 40, 45 and 50 vol. % of zirconium diboride ($ZrB_2$) powders with silicon carbide (SiC) matrix. The SiC-$ZrB_2$ composites and the sintered compacts were produced through spark plasma sintering (SPS) under argon atmosphere, and its physical, electrical, and mechanical properties were examined. Also, the thermal image analysis of the SiC-$ZrB_2$ composites was examined. Reactions between $\beta$-SiC and $ZrB_2$ were not observed via x-ray diffraction (XRD) analysis. The apparent porosity of the SiC+30vol.%$ZrB_2$, SiC+35vol.%$ZrB_2$, SiC+40vol.%$ZrB_2$, SiC+45vol.%$ZrB_2$ and SiC+50vol.%$ZrB_2$ composites were 7.2546, 0.8920, 0.6038, 1.0981, and 10.0108%, respectively. The XRD phase analysis of the sintered compacts demonstrated a high phase of SiC and $ZrB_2$. Among the $SiC+ZrB_2$ composites, the SiC+50vol.%$ZrB_2$ composite had the lowest flexural strength, 290.54MPa, the other composites had more than 980MPa flexural strength except the SiC+30vol.%$ZrB_2$ composite; the SiC+40vol.%$ZrB_2$ composite had the highest flexural strength, 1011.34MPa, at room temperature. The electrical properties of the SiC-$ZrB_2$ composites had positive temperature coefficient resistance (PTCR). The V-I characteristics of the SiC-$ZrB_2$ composites had a linear shape in the temperature range from room to $500^{\circ}C$. The electrical resistivities of the SiC+30vol.%$ZrB_2$, SiC+35vol.%$ZrB_2$, SiC+40vol.%$ZrB_2$ SiC+45vol.%$ZrB_2$ and SiC+50vol.%$ZrB_2$ composites were $4.573\times10^{-3}$, $1.554\times10^{-3}$, $9.365\times10^{-4}$, $6.999\times10^{-4}$, and $6.069\times10^{-4}\Omega{\cdot}cm$, respectively, at room temperature, and their resistance temperature coefficients were $1.896\times10^{-3}$, $3.064\times10^{-3}$, $3.169\times10^{-3}$, $3.097\times10^{-3}$, and $3.418\times10^{-3}/^{\circ}C$ in the temperature range from room to $500^{\circ}C$, respectively. Therefore, it is considered that among the sintered compacts the SiC+35vol.%$ZrB_2$, SiC+40vol.%$ZrB_2$ and SiC+45vol.%$ZrB_2$ composites containing the most outstanding mechanical properties as well as PTCR and V-I characteristics can be used as an energy friendly ceramic heater or ohmic-contact electrode material through SPS.

• 한국균학회소식:학술대회논문집
• /
• 2015.05a
• /
• pp.53-53
• /
• 2015

Agrobacterium tumefaciens-mediated transformation(ATMT) of Flammulina velutipes was used to produce a diverse number of transformants to discover the functions of gene that is vital for its variation color, spore pattern and cellulolytic activity. Futhermore, the transformant pool will be used as a good genetic resource for studying gene functions. Agrobacterium-mediated transformation was conducted in order to generate intentional mutants of F. velutipes strain KACC42777. Then Agrobacterium tumefaciens AGL-1 harboring pBGgHg was transformed into F. velutipes. This method is use to determine the functional gene of F. velutipes. Inverse PCR was used to insert T-DNA into the tagged chromosomal DNA segments and conducting sequence analysis of the F. velutipes. But this experiment had trouble in diverse morphological mutants because of dikaryotic nature of mushroom. It needed to make monokaryotic fruiting varients which introduced genes of compatible mating types. In this study, next generation sequencing data was generated from 28 strains of Flammulina velutipes with different phenotypes using Illumina Hiseq platform. Filtered short reads were initially aligned to the reference genome (KACC42780) to construct a SNP matrix. And then we built a phylogenetic tree based on the validated SNPs. The inferred tree represented that white- and brown- fruitbody forming strains were generally separated although three brown strains, 4103, 4028, and 4195, were grouped with white ones. This topological relationship was consistently reappeared even when we used randomly selected SNPs. Group I containing 4062, 4148, and 4195 strains and group II containing 4188, 4190, and 4194 strains formed early-divergent lineages with robust nodal supports, suggesting that they are independent groups from the members in main clades. To elucidate the distinction between white-fruitbody forming strains isolated from Korea and Japan, phylogenetic analysis was performed using their SNP data with group I members as outgroup. However, no significant genetic variation was noticed in this study. A total of 28 strains of Flammulina velutipes were analyzed to identify the genomic regions responsible for producing white-fruiting body. NGS data was yielded by using Illumina Hiseq platform. Short reads were filtered by quality score and read length were mapped on the reference genome (KACC42780). Between the white- and brown fruitbody forming strains. There is a high possibility that SNPs can be detected among the white strains as homozygous because white phenotype is recessive in F. velutipes. Thus, we constructed SNP matrix within 8 white strains. SNPs discovered between mono3 and mono19, the parental monokaryotic strains of 4210 strain (white), were excluded from the candidate. If the genotypes of SNPs detected between white and brown strains were identical with those in mono3 and mono19 strains, they were included in candidate as a priority. As a result, if more than 5 candidates SNPs were localized in single gene, we regarded as they are possibly related to the white color. In F. velutipes genome, chr01, chr04, chr07,chr11 regions were identified to be associated with white fruitbody forming. White and Brown Fruitbody strains can be used as an identification marker for F. veluipes. We can develop some molecular markers to identify colored strains and discriminate national white varieties against Japanese ones.

• The korean journal of orthodontics
• /
• v.27 no.5 s.64
• /
• pp.815-824
• /
• 1997

The purpose of this study is to develop the basic algorithm for the finite element method modeling of individual malocclusions. Usually, a great deal of time is spent in preprocessing. To reduce the time required, we developed a standardized procedure for measuring the position of each tooth and a program to automatically preprocess. The following procedures were carried to complete this study. 1. Twenty-eight teeth morphologies were constructed three-dimensionally for the finite element analysis and saved as separate files. 2. Standard brackets were attached so that the FA points coincide with the center of the brackets. 3. The study model of a patient was made. 4. Using the study model, the crown inclination, angulation, and the vertical distance from the tip of a tooth was measured by using specially designed tools. 5. The arch form was determined from a picture of the model with an image processing technique. 6. The measured data were input as a rotational matrix. 7. The program provides an output file containing the necessary information about the three-dimensional position of teeth, which is applicable to several finite element programs commonly used. The program for a basic algorithm was made with Turbo-C and the subsequent outfile was applied to ANSYS. This standardized model measuring procedure and the program reduce the time required, especially for preprocessing and can be applied to other malocclusions easily.

• The Korean Journal of Nuclear Medicine Technology
• /
• v.22 no.1
• /
• pp.23-27
• /
• 2018

Purpose The introduction of bone scan has been reported as a useful tool in the diagnosis, treatment, and treatment response of skeletal disease. The purpose of this study is to improve the anatomical information and tolerance of the bone by combining bone scan and pelvic X-ray without additional radiation exposure. Materials and Methods From November 2015 to August 2016, 236 patients(64 men and 172 women, average age $50.96{\pm}15.39years$) take Bone scan and Pelvis AP(Anteroposterior) X-ray scan at the National Cancer Center. The scan equipment was a gamma camera, Symbia Ecam (SIEMENS, Germany), and a digital x-ray, DRS-800 (Listem, Korea). Osirix version 3.8.1 (Osirix, USA) and Stata/SE version 14.0 (StataCorp, USA) were used for image combination and analysis. The patient was intravenously injected with $^{99m}Tc-DPD$ (740 MBq), and the scan was performed 2 to 4 hours later. Gamma camera image acquisition were Matrix size $256{\times}1024$, Zoom 1.00, and scan speed 17 cm/min. The digital X-ray was made with a collimator size of $14^{{\prime}{\prime}}{\times}17^{{\prime}{\prime}}$, 77 kVp (60 to 97 kVp) and an average of 30 mAs (20 to 48). ASIS and pubic symphysis Select virtual points then Combine three virtual points and pelvic contour lines. The acquired images were evaluated by three radiologists who worked for more than 5 years in the nuclear medicine department. Results Of the total 236 patients, 216 (91.53%) were matched. The median and range (min~max) of the age were 67 (46~81) years old in the unmatched group and 52 (22~87) years old in the matched group, The Wilcoxon rank-sum test was performed to determine whether age was different between the two groups. As a result, the age difference between the two groups was statistically significant at p < 0.0001. Of the 64 men, 60 (93.75%) were match and of the 172 women, 156 (93.75%) were match. There was no statistically significant difference according to gender(p = 0.4542). Of the 54 patients without pelvic lesions, 54 (100.00%) were match, and 162 (89.01%) of 182 patients with pelvic lesions were match. There was a statistically significant difference according to the presence of pelvic lesions. Conclusion There are many variables in the combination of bone scan and pelvic X-ray imaging, and the patient's age and pelvic lesion may have some effect on the image combination. This study is expected to be useful for the diagnosis of pelvic osteosarcoma of children without radiation exposure. It is expected that this combination of images will help to develop the nuclear medicine image.