• Proceedings of the Korea Information Processing Society Conference
• /
• 2023.05a
• /
• pp.510-512
• /
• 2023

In order to enhance a model's capability for detecting facial expressions, this research suggests a pipeline that makes use of the GradCAM component. The patching module and the pseudo-labeling module make up the pipeline. The patching component takes the original face image and divides it into four equal parts. These parts are then each input into a 2Dconvolutional layer to produce a feature vector. Each picture segment is assigned a weight token using GradCAM in the pseudo-labeling module, and this token is then merged with the feature vector using principal component analysis. A convolutional neural network based on transfer learning technique is then utilized to extract the deep features. This technique applied on a public dataset MMI and achieved a validation accuracy of 96.06% which is showing the effectiveness of our method.

• Journal of Intelligence and Information Systems
• /
• v.25 no.3
• /
• pp.161-177
• /
• 2019

In this paper, we study the performance improvement of the answer extraction in Question-Answering system by using sentence dependency parsing result. The Question-Answering (QA) system consists of query analysis, which is a method of analyzing the user's query, and answer extraction, which is a method to extract appropriate answers in the document. And various studies have been conducted on two methods. In order to improve the performance of answer extraction, it is necessary to accurately reflect the grammatical information of sentences. In Korean, because word order structure is free and omission of sentence components is frequent, dependency parsing is a good way to analyze Korean syntax. Therefore, in this study, we improved the performance of the answer extraction by adding the features generated by dependency parsing analysis to the inputs of the answer extraction model (Bidirectional LSTM-CRF). The process of generating the dependency graph embedding consists of the steps of generating the dependency graph from the dependency parsing result and learning the embedding of the graph. In this study, we compared the performance of the answer extraction model when inputting basic word features generated without the dependency parsing and the performance of the model when inputting the addition of the Eojeol tag feature and dependency graph embedding feature. Since dependency parsing is performed on a basic unit of an Eojeol, which is a component of sentences separated by a space, the tag information of the Eojeol can be obtained as a result of the dependency parsing. The Eojeol tag feature means the tag information of the Eojeol. The process of generating the dependency graph embedding consists of the steps of generating the dependency graph from the dependency parsing result and learning the embedding of the graph. From the dependency parsing result, a graph is generated from the Eojeol to the node, the dependency between the Eojeol to the edge, and the Eojeol tag to the node label. In this process, an undirected graph is generated or a directed graph is generated according to whether or not the dependency relation direction is considered. To obtain the embedding of the graph, we used Graph2Vec, which is a method of finding the embedding of the graph by the subgraphs constituting a graph. We can specify the maximum path length between nodes in the process of finding subgraphs of a graph. If the maximum path length between nodes is 1, graph embedding is generated only by direct dependency between Eojeol, and graph embedding is generated including indirect dependencies as the maximum path length between nodes becomes larger. In the experiment, the maximum path length between nodes is adjusted differently from 1 to 3 depending on whether direction of dependency is considered or not, and the performance of answer extraction is measured. Experimental results show that both Eojeol tag feature and dependency graph embedding feature improve the performance of answer extraction. In particular, considering the direction of the dependency relation and extracting the dependency graph generated with the maximum path length of 1 in the subgraph extraction process in Graph2Vec as the input of the model, the highest answer extraction performance was shown. As a result of these experiments, we concluded that it is better to take into account the direction of dependence and to consider only the direct connection rather than the indirect dependence between the words. The significance of this study is as follows. First, we improved the performance of answer extraction by adding features using dependency parsing results, taking into account the characteristics of Korean, which is free of word order structure and omission of sentence components. Second, we generated feature of dependency parsing result by learning - based graph embedding method without defining the pattern of dependency between Eojeol. Future research directions are as follows. In this study, the features generated as a result of the dependency parsing are applied only to the answer extraction model in order to grasp the meaning. However, in the future, if the performance is confirmed by applying the features to various natural language processing models such as sentiment analysis or name entity recognition, the validity of the features can be verified more accurately.

• Korean Journal of Clinical Laboratory Science
• /
• v.55 no.4
• /
• pp.261-268
• /
• 2023

[¹⁸F]Fluorocholine is a radiopharmaceutical used non-invasively in positron emission tomography to diagnose parathyroid adenoma, prostate cancer, and hepatocellular carcinoma by evaluating the choline metabolism. In this study, a radiolabeling method for [¹⁸F]fluorocholine was optimized using a solid phase extraction (SPE) cartridge. [¹⁸F]Fluorocholine was labeled in two steps using an automated synthesizer. In the first step, dibromomethane was reacted with [¹⁸F]KF/K2.2.2/K₂CO₃ to obtain the intermediate [¹⁸F]fluorobromomethane. In the second step, [¹⁸F]fluorobromomethane was passed through a Sep-Pak^Ⓡ Silica SPE cartridge to remove the impurities and then reacted with N,N-dimethylaminoethanol (DMAE) in a Sep-Pak^Ⓡ C18 SPE cartridge to label [¹⁸F]fluorocholine. The reaction conditions of [¹⁸F]fluorocholine were optimized. The synthesis yield was confirmed according to the number of silica cartridges and DMAE concentration. No statistically significant difference in the synthesis yield of [¹⁸F]fluorocholine was observed when using four or three silica cartridges (P>0.05). The labeling yield was 11.5±0.5% (N=4) when DMAE was used as its original solution. On the other hand, when diluted to 10% with dimethyl sulfoxide, the radiochemical yield increased significantly to 30.1±5.2% (N=20). In conclusion, [¹⁸F]Fluorocholine for clinical use can be synthesized stably in high yield by applying an optimized synthesis method.

• Journal of Plant Biotechnology
• /
• v.48 no.3
• /
• pp.165-172
• /
• 2021

The solubilization of isolated proteins into the adequate buffer containing of surfactants is primary step for proteomic analysis. Particularly, sodium dodecyl sulfate (SDS) is the most widely used surfactant, however, it is not compatible with mass spectrometry (MS). Therefore, it must be removed prior to MS analysis through rigorous washing, which eventually results in inevitable protein loss. Recently, photocleavable surfactant, 4-hexylphenylazosulfonate (Azo), was reported which can be easily degraded by UV irradiation and is compatible with MS during proteomic approach using animal tissues. In this study, we employed comparative label-free proteomic analysis for evaluating the solubilization efficacies of the Azo and SDS surfactants using rice leave proteins. This approach led to identification of 3,365 proteins of which 682 proteins were determined as significantly modulated. Further, according to the subcellular localization prediction in SDS and Azo, proteins localized in the chloroplast were the major organelle accounting for 64% of the total organelle in the SDS sample, while only 37.5% of organelle proteins solubilized in the Azo were predicted to be localized in chloroplast. Taken together, this study validates the efficient solubilization of total protein isolated from plant material for bottom-up proteomics. Azo surfactant is suitable as substitute of SDS and promising for bottom-up proteomics as it facilitates robust protein extraction, rapid washing step during enzymatic digestion, and MS analysis.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.19 no.1
• /
• pp.161-170
• /
• 2019

Multi-product mass production small and medium enterprise factories have a wide variety of products and a large number of products, wasting manpower and expenses for inventory management. In addition, there is no way to check the status of inventory in real time, and it is suffering economic damage due to excess inventory and shortage of stock. There are many ways to build a real-time data collection environment, but most of them are difficult to afford for small and medium-sized companies. Therefore, smart factories of small and medium enterprises are faced with difficult reality and it is hard to find appropriate countermeasures. In this paper, we implemented the contents of extension of existing inventory management method through character extraction on label with barcode and QR code, which are widely adopted as current product management technology, and evaluated the effect. Technically, through preprocessing using OpenCV for automatic recognition and classification of stock labels and barcodes, which is a method for managing input and output of existing products through computer image processing, and OCR (Optical Character Recognition) function of Google vision API. And it is designed to recognize the barcode through Zbar. We propose a method to manage inventory by real-time image recognition through Raspberry Pi without using expensive equipment.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.18 no.7
• /
• pp.935-943
• /
• 1993

In this paper, robust recognition system for tilted facial image was developed. At first, standard facial image and lilted facial image are captured by CCTV camera and then transformed into binary image. The binary image is processed in order to obtain contour image by Laplacian edge operator. We trace and delete outermost edge line and use inner contour lines. We label four inner contour lines in order among the inner lines, and then we extract left and right eye with known distance relationship and with two eyes coordinates, and calculate slope information. At last, we rotate the tilted image in accordance with slope information and then calculate the ten distance features between element and element. In order to make the system invariant to image scale, we normalize these features with distance between left and righ eye. Experimental results show 88% recognition rate for twenty five face images when tilted degree is considered and 60% recognition rate when tilted degree is not considered.

• Proceedings of the Korean Vacuum Society Conference
• /
• 2016.02a
• /
• pp.384.1-384.1
• /
• 2016

We reported that amorphous silicon (a-Si) thin film provide sample plate exhibiting a multimodality to measure biomolecules by secondary ion mass spectrometry (SIMS) and laser desorption/ionization mass spectrometry (LDI-MS). Kim et al.1 reported that a-Si thin film were suitable to detect small molecules such as drugs and peptides by SIMS and LDI-MS. Recently, bacterial identification has been required in many fields such as food analysis, veterinary science, ecology, agriculture, and so on.2 Mass spectrometry is emerging for identifying and profiling microbiology samples from its advantageous characters of label-free and shot-time analysis. Five species of bacteria - S. aureus, G. glutamicum, B. kurstaki, B. sphaericus, and B. licheniformis - were sampled for MS analysis without lipid extraction in sample preparation steps. The samples were loaded onto the a-Si thin film with a thickness of 100 nm which did not only considered laser-beam penetration but also surface homogeneity. Mass spectra were recorded in both positive and negative ionization modes for more analytical information. High reproducibility and sensitivity of mass spectra were demonstrated in a mass range up to mass-to-charge ratio(m/z) 1200 by applying the a-Si thin film in mentioned above MS. Principle component analysis (PCA) - a popular statistical analysis widely used in data processing was employed to differentiate between five bacterial species. The PCA results verified that each bacterial species were readily distinguished and differentiated effectively from our MS approach. It shows a new opportunity to rapid bacterial profiling and identification in clinical microbiology. More details will be discussed in the presentation.

• Journal of Institute of Control, Robotics and Systems
• /
• v.20 no.8
• /
• pp.868-874
• /
• 2014

This paper proposes a novel mapping algorithm in Omni-directional Vision SLAM based on an obstacle's feature extraction using Lucas-Kanade Optical Flow motion detection and images obtained through fish-eye lenses mounted on robots. Omni-directional image sensors have distortion problems because they use a fish-eye lens or mirror, but it is possible in real time image processing for mobile robots because it measured all information around the robot at one time. In previous Omni-Directional Vision SLAM research, feature points in corrected fisheye images were used but the proposed algorithm corrected only the feature point of the obstacle. We obtained faster processing than previous systems through this process. The core of the proposed algorithm may be summarized as follows: First, we capture instantaneous $360^{\circ}$ panoramic images around a robot through fish-eye lenses which are mounted in the bottom direction. Second, we remove the feature points of the floor surface using a histogram filter, and label the candidates of the obstacle extracted. Third, we estimate the location of obstacles based on motion vectors using LKOF. Finally, it estimates the robot position using an Extended Kalman Filter based on the obstacle position obtained by LKOF and creates a map. We will confirm the reliability of the mapping algorithm using motion estimation based on fisheye images through the comparison between maps obtained using the proposed algorithm and real maps.

• Preventive Nutrition and Food Science
• /
• v.8 no.4
• /
• pp.301-305
• /
• 2003

This study was conducted to develop an HPLC method for determining vitamin B$_{12}$ in fortified foods which has typically been determined by microbiological assays according to AOAC and Korean Food Code approved methods. Vitamin B$_{12}$ (cyanocobalamin) was determined by reversed-phase HPLC with a triple column and UV/VIS dectector (550 nm) using the column switching technique after extraction with 5 mM potassium phosphate solution by sonication without a clean-up procedure. The recovery of spiked samples and limit of detection (LOD) by HPLC were 78.6 ∼107.5 % and 2 ppb ($\mu\textrm{g}$/kg), respectively. The LOD of the microbiological assay (MBA) was much lower than that of HPLC. The concentrations of vitamin B$_{12}$ analyzed in all tested samples (n=12) confirmed compliance with declared label claims. The range of recovery ratio by the HPLC method when compared to the microbiological assay was 76.2 ∼140.0 %. There was not significant difference between the HPLC and MBA methods (p < 0.01) with r=0.9791 and linear regression y=0.9923x-0.04. The HPLC method for determining vitamin B$_{12}$ using the column-switching technique appears to be suitable for determining vitamin B$_{12}$ concentrations above 1 $\mu\textrm{g}$/100 g in fortified foods.ied foods.

• Journal of Information Processing Systems
• /
• v.15 no.6
• /
• pp.1306-1325
• /
• 2019

As the world is moving towards digitization, data is generated from various sources at a faster rate. It is getting humungous and is termed as big data. The financial sector is one domain which needs to leverage the big data being generated to identify financial risks, fraudulent activities, and so on. The design of predictive models for such financial big data is imperative for maintaining the health of the country's economics. Financial data has many features such as transaction history, repayment data, purchase data, investment data, and so on. The main problem in predictive algorithm is finding the right subset of representative features from which the predictive model can be constructed for a particular task. This paper proposes a correlation-based method using submodular optimization for selecting the optimum number of features and thereby, reducing the dimensions of the data for faster and better prediction. The important proposition is that the optimal feature subset should contain features having high correlation with the class label, but should not correlate with each other in the subset. Experiments are conducted to understand the effect of the various subsets on different classification algorithms for loan data. The IBM Bluemix BigData platform is used for experimentation along with the Spark notebook. The results indicate that the proposed approach achieves considerable accuracy with optimal subsets in significantly less execution time. The algorithm is also compared with the existing feature selection and extraction algorithms.