• Journal of the Korea Society of Computer and Information
• /
• v.28 no.4
• /
• pp.21-29
• /
• 2023

In this paper, we propose an AI-based system for automatically classifying industry and occupation codes in the population census. The accurate classification of industry and occupation codes is crucial for informing policy decisions, allocating resources, and conducting research. However, this task has traditionally been performed by human coders, which is time-consuming, resource-intensive, and prone to errors. Our system represents a significant improvement over the existing rule-based system used by the statistics agency, which relies on user-entered data for code classification. In this paper, we trained and evaluated several models, and developed an ensemble model that achieved an 86.76% match accuracy in industry and 81.84% in occupation, outperforming the best individual model. Additionally, we propose process improvement work based on the classification probability results of the model. Our proposed method utilizes an ensemble model that combines transfer learning techniques with pre-trained models. In this paper, we demonstrate the potential for AI-based systems to improve the accuracy and efficiency of population census data classification. By automating this process with AI, we can achieve more accurate and consistent results while reducing the workload on agency staff.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.18 no.1
• /
• pp.287-294
• /
• 2017

BIM technology has been used in the domestic AEC field since the middle 2000s. BIM has proved its worth in cutting-edge buildings, mega-buildings and freeform buildings in particular. Many freeform buildings could not be completed due to the low level of construction technique. However, many successful cases emerged after adopting digital technology, including BIM which encouraged architects to challenge freeform designs. The modeling software that can generate the freeform shape are not usually able to build the efficient BIM data type in the AEC industry. In this study a process model of the parametric freeform construction member generation and conversion to BIM data is shown and the prototype system is demonstrated.

• Korean Chemical Engineering Research
• /
• v.48 no.1
• /
• pp.10-15
• /
• 2010

Intensive works have been carried out to develop more efficient solid catalysts for biodiesel production from various feedstocks including refined oils and waste fats. Among many catalysts, metal oxides and ion exchange resins are the most intensively studied ones. With regard to metal oxide catalysts, major research activities have focused on the identification of the active compounds and their immobilizing methods on the supports. As metal oxide catalysts have strong thermal stability, they may be used in simultaneous transesterification and esterification of waste fats. However, ion exchange resin catalysts were mainly applied in the esterification of the free fatty acids in waste fats because of their lower thermal stability. For both solid catalysts, further works are needed to make them to be used in commercial process. Especially fast deactivation of the solid catalyst would be the most challenging problem.

• The Journal of the Korea institute of electronic communication sciences
• /
• v.9 no.12
• /
• pp.1337-1344
• /
• 2014

The objective of an echo canceller connected to any end of a communication line such as digital subscriber line (DSL) is to compensate the outgoing transmit signal in the receiving path that the hybrid circuit leaks. The echo canceller working in a full duplex environment is an adaptive system driven by the local signal. Conventional echo canceller that implement the least mean square (LMS) algorithm provides a low computational burden but poor convergence properties. The length of the echo canceller will directly affect both the degree of performance and the convergence speed of the adaptation process. To cancel long time-varying echoes, the number of tap coefficients of a conventional echo canceller must be large, which decreases the convergence speed of the adaptive filter. This paper proposes an alternative technique for the echo cancellation in a telecommunication channel. The new technique employs the bi-directional least mean square (LMS) algorithm for adaptively computing the optimal set of the coefficients of the echo canceller, which is composed of weighted combination of both feedforward and feedback algorithms. Finally, Simulation results as well as mathematical analysis demonstrates that the proposed echo canceller has faster convergence speed than the conventional LMS echo canceller with nearly equivalent complexity of computation.

• The Transactions of the Korea Information Processing Society
• /
• v.3 no.6
• /
• pp.1598-1606
• /
• 1996

In order to achieve high accuracy of off-line character recognition(OCR) systems, the recognized text must be processed through a post-processing stage using contextual information. In this paper, we reclassify Korean word classes in terms of OCR word correction. And we collect combinations of Korean particles(approximately 900) linguistic verbal from(around 800). We aggregate 9 Korean irregular verbal phrases defined from a Korean linguistic point of view. Using these Korean word information and a Head-tail method, we can correct misrecognized words. A Korean character recognizer demonstrates 93.7% correct character recognition without a post-processing stage. The entire recognition rate of our system with a post-processing stage exceeds 97% correct character recognition.

• The KIPS Transactions:PartB
• /
• v.16B no.1
• /
• pp.79-84
• /
• 2009

This paper explores a method of text watermarking for agglutinative languages and develops a syntactic tree-based syntactic constituent movement scheme. Agglutinative languages provide a good ground for the syntactic tree-based natural language watermarking because syntactic constituent order is relatively free. Our proposed natural language watermarking method consists of seven procedures. First, we construct a syntactic dependency tree of unmarked text. Next, we perform clausal segmentation from the syntactic tree. Third, we choose target syntactic constituents, which will move within its clause. Fourth, we determine the movement direction of the target constituents. Then, we embed a watermark bit for each target constituent. Sixth, if the watermark bit does not coincide with the direction of the target constituent movement, we displace the target constituent in the syntactic tree. Finally, from the modified syntactic tree, we obtain a marked text. From the experimental results, we show that the coverage of our method is 91.53%, and the rate of unnatural sentences of marked text is 23.16%, which is better than that of previous systems. Experimental results also show that the marked text keeps the same style, and it has the same information without semantic distortion.

• Journal of KIISE:Computer Systems and Theory
• /
• v.30 no.7_8
• /
• pp.359-369
• /
• 2003

We present efficient generalized algorithms for all-to-all personalized communication operations in a 2D torus. All-to-all personalized communication, or complete exchange, is at the heart of numerous applications, such as matrix transposition, Fast Fourier Transform(FFT), and distributed table lookup. Some algorithms have been Presented when the number of nodes is power-of-2 or multiple-of-four form, but there has been no result for general cases yet. We first present complete exchange algorithm called multiple-Hop-2D when the number of nodes is in the form of multiple-of-two. Then by extending this algorithm, we present two algorithms for an arbitrary number of nodes. Split-and-Merge algorithm first splits the whole network into zones. After each zone performs complete exchange, merge is applied to finish the desired complete exchange. By handling extra steps in Double-Hop-2D algorithm, Modified Double-Hop-2D algorithm performs complete exchange operation for general cases. Finally, we compare the required start-up time for these algorithms.

• Korean Journal of Mineralogy and Petrology
• /
• v.34 no.2
• /
• pp.95-106
• /
• 2021

The SO₄ in the jarosite structure can be substituted by other oxyanions, and therefore, the transition of jarosite to goethite plays a very important role in controlling the behavior of oxyanions. In this study, the phase change according to the species of the oxyanion in jarosite and the related behavior of the oxyanion was studied by mineralogical and geochemical methods when jarosite, which is coprecipitated with various oxynions, undergoes a phase change by a reductive dissolution. Jarosite substituted by five oxyanions by 5 mol% was used in this study. The mineral phase change induced by reductive dissolution using ammonium oxalate was investigated, and the order of phase transition rate of jarosite to goethite was MoO₄-jarosite ≥ SeO₄-jarosite ≥ CrO₄-jarosite > pure jarosite > SeO₃-jarosite > AsO₄-jarosite, showing that the transition rates vary depending on the substituted oxyanion. The resultant concentration of the leached Fe was slightly different depending on the type of oxyanion and time but did not show a noticeable difference. The concentration of each oxyanion leached according to the change of the mineral phase showed that the order of concentration of oxyanions was Mo > Se(SeO₃) > As > Se(SeO₄) > Cr in general, and showed a slight increase with time. This trend was related to the species of oxyanions rather than mineral phase change. The results of this study showed that the phase transition of jarosite to goethite was affected by the species of oxyanions, but this tendency did not affect the concentrations leached oxyanions.

• Journal of the Mineralogical Society of Korea
• /
• v.32 no.3
• /
• pp.173-184
• /
• 2019

Variations in grain size distribution and clay mineral assemblage are closely related to the sedimentary facies that reflect depositional conditions during the glacial and interglacial periods. Gravity cores BS17-GC15 and BS17-GC04 were collected from the continental shelf and rise in the eastern Bellingshausen Sea during a cruise of the ANA07D Cruise Expedition by the Korea Polar Research Institute in 2017. Core sediments in BS17-GC15 consisted of subglacial diamicton, gravelly muddy sand, and bioturbated diatom-bearing mud from the bottom to the top sediments. Core sediments in BS17-GC04 comprised silty mud with turbidites, brownish structureless mud, laminated mud, and brownish silty bioturbated diatom-bearing mud from the bottom to the top sediments. The clay mineral assemblages in the two core sediments mainly consisted of smectite, chlorite, illite, and kaolinite. The clay mineral contents in core GC15 showed a variation in illite from 28.4 % to 44.5 % in down-core changes. Smectite contents varied from 31.1 % in the glacial period to 20 % in the deglacial period and 25.1 % in the interglacial period. Chlorite and kaolinite contents decreased from 40.5 % in the glacial period to 30.3 % in the interglacial period. The high contents of illite and chlorite indicated a terrigenous detritus supply from the bedrocks of the Antarctic Peninsula. Core GC04 from the continental rise showed a decrease in the average smectite content from 47.2 % in the glacial period to 20.6 % in the interglacial period, while the illite contents increased from the 21.3 % to 43.2 % from the glacial to the interglacial period. The high smectite contents in core GC04 during the glacial period may be supplied from Peter I Island, which has a known smectite-rich sediment contributed by Antarctic Circumpolar Currents. Conversely, the decrease in smectite and increase in chlorite and illite contents during the interglacial period was likely caused by a higher supply of chlorite- and illite-enriched sediment from the eastern Bellingshausen Sea shelf by the southwestward flowing contour current.

• Journal of Broadcast Engineering
• /
• v.27 no.2
• /
• pp.228-239
• /
• 2022

In this paper, we propose an unmanned vehicle scratch detection deep learning model for car sharing services. Conventional scratch detection models consist of two steps: 1) a deep learning module for scratch detection of images before and after rental, 2) a manual matching process for finding newly generated scratches. In order to build a fully automatic scratch detection model, we propose a one-step unmanned scratch detection deep learning model. The proposed model is implemented by applying transfer learning and fine-tuning to the deep learning model that detects changes in satellite images. In the proposed car sharing service, specular reflection greatly affects the scratch detection performance since the brightness of the gloss-treated automobile surface is anisotropic and a non-expert user takes a picture with a general camera. In order to reduce detection errors caused by specular reflected light, we propose a preprocessing process for removing specular reflection components. For data taken by mobile phone cameras, the proposed system can provide high matching performance subjectively and objectively. The scores for change detection metrics such as precision, recall, F1, and kappa are 67.90%, 74.56%, 71.08%, and 70.18%, respectively.