• Journal of Intelligence and Information Systems
• /
• v.26 no.2
• /
• pp.79-104
• /
• 2020

Recently, as deep learning has attracted attention, the use of deep learning is being considered as a method for solving problems in various fields. In particular, deep learning is known to have excellent performance when applied to applying unstructured data such as text, sound and images, and many studies have proven its effectiveness. Owing to the remarkable development of text and image deep learning technology, interests in image captioning technology and its application is rapidly increasing. Image captioning is a technique that automatically generates relevant captions for a given image by handling both image comprehension and text generation simultaneously. In spite of the high entry barrier of image captioning that analysts should be able to process both image and text data, image captioning has established itself as one of the key fields in the A.I. research owing to its various applicability. In addition, many researches have been conducted to improve the performance of image captioning in various aspects. Recent researches attempt to create advanced captions that can not only describe an image accurately, but also convey the information contained in the image more sophisticatedly. Despite many recent efforts to improve the performance of image captioning, it is difficult to find any researches to interpret images from the perspective of domain experts in each field not from the perspective of the general public. Even for the same image, the part of interests may differ according to the professional field of the person who has encountered the image. Moreover, the way of interpreting and expressing the image also differs according to the level of expertise. The public tends to recognize the image from a holistic and general perspective, that is, from the perspective of identifying the image's constituent objects and their relationships. On the contrary, the domain experts tend to recognize the image by focusing on some specific elements necessary to interpret the given image based on their expertise. It implies that meaningful parts of an image are mutually different depending on viewers' perspective even for the same image. So, image captioning needs to implement this phenomenon. Therefore, in this study, we propose a method to generate captions specialized in each domain for the image by utilizing the expertise of experts in the corresponding domain. Specifically, after performing pre-training on a large amount of general data, the expertise in the field is transplanted through transfer-learning with a small amount of expertise data. However, simple adaption of transfer learning using expertise data may invoke another type of problems. Simultaneous learning with captions of various characteristics may invoke so-called 'inter-observation interference' problem, which make it difficult to perform pure learning of each characteristic point of view. For learning with vast amount of data, most of this interference is self-purified and has little impact on learning results. On the contrary, in the case of fine-tuning where learning is performed on a small amount of data, the impact of such interference on learning can be relatively large. To solve this problem, therefore, we propose a novel 'Character-Independent Transfer-learning' that performs transfer learning independently for each character. In order to confirm the feasibility of the proposed methodology, we performed experiments utilizing the results of pre-training on MSCOCO dataset which is comprised of 120,000 images and about 600,000 general captions. Additionally, according to the advice of an art therapist, about 300 pairs of 'image / expertise captions' were created, and the data was used for the experiments of expertise transplantation. As a result of the experiment, it was confirmed that the caption generated according to the proposed methodology generates captions from the perspective of implanted expertise whereas the caption generated through learning on general data contains a number of contents irrelevant to expertise interpretation. In this paper, we propose a novel approach of specialized image interpretation. To achieve this goal, we present a method to use transfer learning and generate captions specialized in the specific domain. In the future, by applying the proposed methodology to expertise transplant in various fields, we expected that many researches will be actively conducted to solve the problem of lack of expertise data and to improve performance of image captioning.

• Journal of the Korea Concrete Institute
• /
• v.16 no.4 s.82
• /
• pp.477-484
• /
• 2004

As a part of durability improvement of concrete-structure, penetrating water repellency of liquid type is applied to concrete surface. Besides, a related standard is made recently, but the standard has been prescribe for initial settlement state of penetrating water repellency of liquid type, to the exclusion of performance variation depending time and outdoor environment factor. For measurement of performance variation, we measured the weight of outdoor exposure specimen every regular intervals and check a measured value against a measured value of different condition specimen. Moreover, after the application of penetrating water repellent, measured a adhesive strength in tension between cement-polymer modified waterproof coatings and surface of specimen. The applied penetrating water repellent is a emulsified silicon type with a deep penetration depth. In view of the results so far achieved, the more a Quantity of application and active solid content does get, the deeper penetrating water repellency of emulsion type Penetrate get longer and supplied moisture increase in quantity, a penetrating water repellency of liquid type penetrates more deep, but a quantity of water absorption increase gradually. Perhaps this result is caused by a reduction of active solid content on concrete surface, because active ingredient is moved into the concrete by dissolution.

• The Journal of The Korea Institute of Intelligent Transport Systems
• /
• v.6 no.2
• /
• pp.138-145
• /
• 2007

Luminous traffic safety mark is restricted to use only the place that has a thick fog, many night traffic accidents, limited field of view due to structure of road. Recently, LEDs are used for luminous traffic safety mark, but we propose an organic LED for a novel luminous traffic safety mark in the near future. The device structure was $ITO/2-TNATA(500{\AA})/{\alpha}-NPD(200{\AA})/DPVBi(300{\AA})/BCP(10{\AA})/Alq_3(200{\AA})/LiF(10{\AA})/Al:Li(1000{\AA})$. The characteristics of the device are most efficient on occasion of using $N_2$ gas plasma treatment. Current density is $240.71mA/cm^2$ luminance $10,550cd/m^2$, and current efficiency 3.53cd/A at an applied voltage of 10V. The maximum EL wavelength of the fabricated blue organic light-emitting device is 456nm. CIE color coordinates are x=0.1449 and y=0.1633, which is similar to NTSC deep-blue color with CIE color coordinates of x=0.14 and y=0.08.

• Proceedings of the Korean Society of Soil and Groundwater Environment Conference
• /
• 2002.04a
• /
• pp.85-92
• /
• 2002

When the United Nation classified as Korea is the one of the water deficit country. The consensus was made that the water is the one of the precious national resources. Government increases their R/D budget trying to get more clean water bodies. For instances, 'Sustainable Water Resources Development' project is the one of major title in '21 Century Frontier Research project and there are several small research projects are undergoing by the Ministry of Agriculture and KARICO. However, when the environmental preservation issue has been get more emphasis, construction of the Surface Dam met the blockage from the environmentalists due to the problem of the their water buried area. Since the most fitting site for surface dam had been used in the past, some engineer move their focus on modification of the existing Dam's height to enlarge its capacity or dredging the bottom of the reservoir recently However dredging evoke water quality problem in return by accumulated materials at the bottom. Last year the Dong Gang Dam plan has been canceled by environmental problem in water buried area of the reservoir. With the point of this view, ground water gets more focus for the one of the useful alternative for clean water bodies. Underground dam technique which had widely applied once in the early nineteen eighties by the KARICO and attenuated due to engineering insufficiency. The technique is newly studied with the advanced engineering technique. Still groundwater usage rate in Korea is much lower comparing with the advanced countries and has many rooms to develop. Wells, under ground dam and radial collector wells are typical facilities up to now. There is little application in Korea for the Recharge Dam, which had been widely used in the advanced countries. The Recharge Dam is technique to conjunct surface water and groundwater body together, This technique had developed to increase groundwater recharge at the beginning This research is the result of the study on the possibility of the development of the new technology, Groundwater Reservoir' which was modified from Recharge Dam. Groundwater Reservoir is like a deep artificial lakes trenched in hard rock aquifer to get groundwater. The advantage of the Groundwater Reservoir is followings 1) It can be developed at the plains area, not in the deep valley 2) Huge water body can be developed without dam 3) Small buried area comparing surface water dam makes the least environmental effect. 4) Trenching cost can be substitute by the income of the selling rock debris 5) Outfit of the reservoir can be modified to match with the site prospect 6) Rock debris can be used as constructing materials 7) It can be used as groundwater recharge system when the heavy rains comes 8) The reservoir looks like scenery lake with huge clean water bodies.

• Economic and Environmental Geology
• /
• v.53 no.2
• /
• pp.213-220
• /
• 2020

When CO₂ foam is injected into the saline aquifer, the relative permeability of CO₂ decreases and its viscosity increases, thereby reducing mobility in porous media and ultimately improving CO₂ storge with enhanced sweep efficiency. In general, surfactants were used to fabricate CO₂ foam. Recently, nanoparticles have been used to form stable foam than surfactant. This paper introduces CO₂ storage technology using nanoparticle stabilized CO₂ foam. If the surface of the hydrophilic nanoparticles is partially modified into a CO₂-philic portion, the particles have an affinity for CO₂ and water, thus forming a stable CO₂ foam even in deep saline aquifers under high temperature and high salinity conditions, thereby it can be stored in the pores of the rock. In terms of economics, injection method using nanopaticle-stabilized CO₂ foam is more expensive than the conventional CO₂ injection, but it is estimated that it will have price competitiveness because the injection efficiency is improved. From an environmental point of view, it is possible to inject chemical substances such as surfactants and nanomaterials into aquifers or reservoirs for specific purposes such as pollutant removal and oil production. However, some studies have shown that nanoparticles and surfactants are toxic to aquatic animals, so environmentally proven substances should be used. Therefore, further research and development will be needed to study the production and injection of nanoparticle-stabilized CO₂ foam that are environmentally safe and economically reasonable.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.20 no.2
• /
• pp.149-155
• /
• 2020

The Korea Highway Traffic Authority provides statistics that analyze the causes of traffic accidents that occurred since 2015 using the Traffic Accident Analysis System (TAAS). it was reported Through TAAS that the driver's forward carelessness was the main cause of traffic accidents in 2018. As statistics on the cause of traffic accidents, 51.2 percent used mobile phones and watched DMB while driving, 14 percent did not secure safe distance, and 3.6 percent violated their duty to protect pedestrians, representing a total of 68.8 percent. In this paper, we propose a system that has improved the advanced driver assistance system ADAS (Advanced Driver Assistance Systems) by utilizing CNN (Convolutional Neural Network) among the algorithms of Deep Learning. The proposed system learns a model that classifies the movement of the driver's face and eyes using Conv2D techniques which are mainly used for Image processing, while recognizing and detecting objects around the vehicle with cameras attached to the front of the vehicle to recognize the driving environment. Then, using the learned visual steering model and driving environment data, the hazard is classified and detected in three stages, depending on the driver's view and driving environment to assist the driver with the forward and blind spots.

• Journal of the Korean Institute of Illuminating and Electrical Installation Engineers
• /
• v.19 no.2
• /
• pp.13-22
• /
• 2005

This research aimed to generate two points of daylighting application as follows: 1) providing daylight performance data with a variety of glazing materials for a large window and 2) designing and evaluating an experimental type of differentiated window. For this purpose, we compared the daylight and distribution performance of new defined type of window configuration to the conventional window counterpart with a variety of glazing materials. The comparison was made for a deep, south-facing perimeter zone with large window, without any interior obstruction. The conventional window is the base single homogeneous glass pane, where as the differentiated window uses of two different glazings; an upper daylight glazing with high visible transmission and lower view window with lower transmittal glass. The daylight performance data was translated into a ratio between outdoor illuminance and the interiors. The simulated analysis of the conventional window indicates that the interior light levels have been changed proportionally dependent on the transmittance of the applied glass. The comparison of daylight distribution analysis showed that the differentiated window has lots of photometric advantage by the optical function of upper daylight window. In particular, the contribution of higher daylight window into deep rear space must be stressed for daylighting application.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.12 no.2
• /
• pp.775-782
• /
• 2011

This study was designed to clarify the morphometrical change of lymph node, deep cortex and lymph follicles in draining lymph nodes of young mice in response to local injection of lipopolysaccharide(LPS). 1. In the group stimulated with LPS, aged 0 day and 3 days, the number of lymph follicles were not significantly different from those of control group. 2. In the group two to four weeks after injection with LPS, aged five days and one week, the number of lymph follicles were significantly increased from those of control group. 3. In the group one to four weeks after injection with LPS, aged 0 day, three days, five days and one week, the area of lymph node and deep cortex increased about 1.5-3 times more than that of the control group. 4. In the group two to four weeks after injection with LPS, aged three days, five days and one week, the lymph follicles(the area: larger than 0.1 mm2) were increased from those of control group. 5. In the group two to four weeks after injection with LPS, aged five days and one week, the lymph follicles(the area: smaller than 0.01 mm2) were increased from those of control group. In view of these experimental findings, the formation of lymph follicles were induced by LPS stimulation from 5 days to one week after birth. The newley formed lymph follicles area in response to LPS may be less than $0.01mm^2$.

• International journal of advanced smart convergence
• /
• v.11 no.1
• /
• pp.19-27
• /
• 2022

Across the world, 'housing' comprises a significant portion of wealth and assets. For this reason, fluctuations in real estate prices are highly sensitive issues to individual households. In Korea, housing prices have steadily increased over the years, and thus many Koreans view the real estate market as an effective channel for their investments. However, if one purchases a real estate property for the purpose of investing, then there are several risks involved when prices begin to fluctuate. The purpose of this study is to design a real estate price 'return rate' prediction model to help mitigate the risks involved with real estate investments and promote reasonable real estate purchases. Various approaches are explored to develop a model capable of predicting real estate prices based on an understanding of the immovability of the real estate market. This study employs the LSTM method, which is based on artificial intelligence and deep learning, to predict real estate prices and validate the model. LSTM networks are based on recurrent neural networks (RNN) but add cell states (which act as a type of conveyer belt) to the hidden states. LSTM networks are able to obtain cell states and hidden states in a recursive manner. Data on the actual trading prices of apartments in autonomous districts between January 2006 and December 2019 are collected from the Actual Trading Price Disclosure System of the Ministry of Land, Infrastructure and Transport (MOLIT). Additionally, basic data on apartments and commercial buildings are collected from the Public Data Portal and Seoul Metropolitan Government's data portal. The collected actual trading price data are scaled to monthly average trading amounts, and each data entry is pre-processed according to address to produce 168 data entries. An LSTM model for return rate prediction is prepared based on a time series dataset where the training period is set as April 2015~August 2017 (29 months), the validation period is set as September 2017~September 2018 (13 months), and the test period is set as December 2018~December 2019 (13 months). The results of the return rate prediction study are as follows. First, the model achieved a prediction similarity level of almost 76%. After collecting time series data and preparing the final prediction model, it was confirmed that 76% of models could be achieved. All in all, the results demonstrate the reliability of the LSTM-based model for return rate prediction.

• Journal of the Korea Society of Computer and Information
• /
• v.27 no.10
• /
• pp.29-41
• /
• 2022

Recently, with the development of deep learning technology, a variety of huge models with excellent performance have been devised by pre-training massive amounts of text data. However, in order for such a model to be applied to real-life services, the inference speed must be fast and the amount of computation must be low, so the technology for model compression is attracting attention. Knowledge distillation, a representative model compression, is attracting attention as it can be used in a variety of ways as a method of transferring the knowledge already learned by the teacher model to a relatively small-sized student model. However, knowledge distillation has a limitation in that it is difficult to solve problems with low similarity to previously learned data because only knowledge necessary for solving a given problem is learned in a teacher model and knowledge distillation to a student model is performed from the same point of view. Therefore, we propose a heterogeneous knowledge distillation method in which the teacher model learns a higher-level concept rather than the knowledge required for the task that the student model needs to solve, and the teacher model distills this knowledge to the student model. In addition, through classification experiments on about 18,000 documents, we confirmed that the heterogeneous knowledge distillation method showed superior performance in all aspects of learning efficiency and accuracy compared to the traditional knowledge distillation.