• Annual Conference of KIPS
• /
• 2018.05a
• /
• pp.377-380
• /
• 2018

인간의 추론 능력이란 문제에 주어진 조건을 보고 문제 해결에 필요한 것이 무엇인지를 논리적으로 생각해 보는 것으로 문제 상황 속에서 일정한 규칙이나 성질을 발견하고 이를 수학적인 방법으로 법칙을 찾아내거나 해결하는 능력을 말한다. 이러한 인간인지 능력과 유사한 인공지능 시스템을 개발하는데 있어서 핵심적 도전은 비구조적 데이터(unstructured data)로부터 그 개체들(object)과 그들간의 관계(relation)에 대해 추론하는 능력을 부여하는 것이라고 할 수 있다. 지금까지 딥러닝(deep learning) 방법은 구조화 되지 않은 데이터로부터 문제를 해결하는 엄청난 진보를 가져왔지만, 명시적으로 개체간의 관계를 고려하지 않고 이를 수행해왔다. 최근 발표된 구조화되지 않은 데이터로부터 복잡한 관계 추론을 수행하는 심층신경망(deep neural networks)은 관계추론(relational reasoning)의 시도를 이해하는데 기대할 만한 접근법을 보여주고 있다. 그 첫 번째는 관계추론을 위한 간단한 신경망 모듈(A simple neural network module for relational reasoning) 인 RN(Relation Networks)이고, 두 번째는 시각적 관찰을 기반으로 실제대상의 미래 상태를 예측하는 범용 목적의 VIN(Visual Interaction Networks)이다. 관계 추론을 수행하는 이들 심층신경망(deep neural networks)은 세상을 객체(objects)와 그들의 관계(their relations)라는 체계로 분해하고, 신경망(neural networks)이 피상적으로는 매우 달라 보이지만 근본적으로는 공통관계를 갖는 장면들에 대하여 객체와 관계라는 새로운 결합(combinations)을 일반화할 수 있는 강력한 추론 능력(powerful ability to reason)을 보유할 수 있다는 것을 보여주고 있다. 본 논문에서는 관계 추론을 수행하는 심층신경망(deep neural networks) 중에서 Sort-of-CLEVR 데이터 셋(dataset)을 사용하여 RN(Relation Networks)의 성능을 재현 및 관찰해 보았으며, 더 나아가 파라미터(parameters) 튜닝을 통하여 RN(Relation Networks) 모델의 성능 개선방법을 제시하여 보았다.

• Journal of Industrial Convergence
• /
• v.17 no.1
• /
• pp.1-6
• /
• 2019

Various cryptographic technologies have been proposed from ancient times and are developing in various ways to ensure the confidentiality of information. Due to exponentially increasing computer power, the encryption key is gradually increasing for security. Technology are being developed; however, security is guaranteed only in a short period of time. With the advent of the 4th Industrial Revolution, encryption technology is required in various fields. Recently, encryption technology using homomorphic encryption has attracted attention. Security threats arise due to the exposure of keys and plain texts used in the decryption processing for the operation of encrypted information. The homomorphic encryption can compute the data of the cipher text and secure process the information without exposing the plain text. When using the homomorphic encryption in processing big data like stored personal information in various services, security threats can be avoided because there is no exposure to key usage and decrypted information.

• Journal of the Korean Society of Food Culture
• /
• v.37 no.5
• /
• pp.385-409
• /
• 2022

This study investigated the trend of research on 'Food science and Nutrition' in previously published papers in the Journal of Korean Society of Food Culture (JKSFC) from 2000 to 2021. Total number of published papers in this category in the JKSFC was 693 which we classified into 7 main categories and 40 subcategories. Of these, 256 articles were on 'experimental cooking' which was the most studied field among 7 main categories. There was a total of 19 published papers under the category of 'microbiology and fermentation'. A total of 133 articles were published on 'functional foods' and provided essential data for discovering new materials under the theme of various physiological active functions of food materials. Furthermore, 107 articles were included in 'food processing and storage', which provided integrated knowledge of economy, stability and practicality based on various technologies. A total of 144 articles was included in the category of 'nutrition'. Under the category of 'nutrition', the most actively studied topic was 'eating behaviors and dietary habits,' and the trending topic was 'use of healthcare big data.' In conclusion, this review would provide trends of various categories of food science and nutrition area for recent 21 years and suggest directions for future research.

• Journal of Information Processing Systems
• /
• v.18 no.3
• /
• pp.359-373
• /
• 2022

The new crown pneumonia (COVID-19) has become a global epidemic. The disease has spread to most countries and poses a challenge to the healthcare system. Contact tracing technology is an effective way for public health to deal with diseases. Many experts have studied traditional contact tracing and developed digital contact tracking. In order to better understand the field of contact tracking, it is necessary to analyze the development of contact tracking in the field of computer science by bibliometrics. The purpose of this research is to use literature statistics and topic analysis to characterize the research literature of contact tracking in the field of computer science, to gain an in-depth understanding of the literature development status of contact tracking and the trend of hot topics over the past decade. In order to achieve the aforementioned goals, we conducted a bibliometric study in this paper. The study uses data collected from the Scopus database. Which contains more than 10,000 articles, including more than 2,000 in the field of computer science. For popular trends, we use VOSviewer for visual analysis. The number of contact tracking documents published annually in the computer field is increasing. At present, there are 200 to 300 papers published in the field of computer science each year, and the number of uncited papers is relatively small. Through the visual analysis of the paper, we found that the hot topic of contact tracking has changed from the past "mathematical model," "biological model," and "algorithm" to the current "digital contact tracking," "privacy," and "mobile application" and other topics. Contact tracking is currently a hot research topic. By selecting the most cited papers, we can display high-quality literature in contact tracking and characterize the development trend of the entire field through topic analysis. This is useful for students and researchers new to field of contact tracking ai well as for presenting our results to other subjects. Especially when comprehensive research cannot be conducted due to time constraints or lack of precise research questions, our research analysis can provide value for it.

• Annual Conference of KIPS
• /
• 2021.05a
• /
• pp.293-296
• /
• 2021

K-water에서는 다목적댐의 관리를 위해 실시간으로 댐수위, 하천 수위 및 강우량 등을 계측하고 있으며, 계측된 값들은 댐을 효과적으로 운영하는데 필요한 데이터로 활용되고 있다. 특히 댐수위 이상 데이터를 탐지하지 못한 채 그대로 사용할 경우 댐의 방류 시기와 방류량 등을 결정하는 중요한 의사결정을 그르칠 수 있으므로 이를 신속히 탐지하는 것이 매우 중요하다. 현재의 자동화된 이상 데이터 탐지방법 중 하나는 현재 데이터가 최댓값과 최솟값을 초과할 때, 다른 하나는 현재 데이터와 일정 시간 동안의 평균값 간의 차이가 관리자가 정한 특정 값을 벗어났을 때를 기준으로 삼고 있다. 전자는 상한과 하한의 초과 여부만 판단하므로 탐지가 쉬우나 정상범위 내에서 발생한 이상 데이터는 탐지가 불가하다. 후자는 관리자의 경험을 통해 판단 조건을 정하기 때문에 객관성이 결여되는 문제가 있다. 특히 방류와 강우가 복합적으로 댐수위에 영향을 미치는 홍수기에 관리자의 경험에 기초한 이상 데이터 판별은 신뢰성의 문제가 있을 수 있다. 따라서 본 연구에서는 기계학습을 최초로 적용하여 이상 데이터를 탐지하고자 하였다. 댐수위, 누적강우량 및 누적방류량 데이터와 댐수위데이터를 가공하여 생성한 댐수위차, 댐수위차평균, 댐수위평균 등 자질들의 다양한 조합을 만든 후 이를 Random Forest, SVM, AdaptiveBoost 및 다층퍼셉트론(MLP) 등과 같은 여러 가지 기계학습모델 등을 통해 이상 데이터를 판별하는 실험(분류)을 하였다. 실험결과 댐수위, 댐수위차, 댐수위-댐수위평균, 누적강우량, 누적방류량 및 댐수위차평균을 사용하였을 때 MLP에서 가장 우수한 성능을 보였다. 이 연구를 통해서 댐수위 이상 데이터를 기계학습의 분류기능을 통해 효과적으로 탐지할 수 있다는 것과 모델의 성능은 실험에 사용한 자질의 수뿐 아니라 자질의 종류에도 큰 영향을 받는다는 것을 알 수 있었다.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2021.10a
• /
• pp.236-239
• /
• 2021

With the growth of AI technology, the scale of Knowledge Graphs continues to be expanded. Knowledge Graphs are mainly expressed as RDF representations that consist of connected triples. Many RDF storages compress and transform RDF triples into the condensed IDs. However, if we try to transform a large scale of RDF triples, it occurs the high processing time and memory overhead because it needs to search the large ID mapping table. In this paper, we propose the method of converting RDF triples using Hash-based ID mapping tables with MapReduce, which is the software framework with a parallel, distributed algorithm. Our proposed method not only transforms RDF triples into Integer-based IDs, but also improves the conversion speed and memory overhead. As a result of our experiment with the proposed method for LUBM, the size of the dataset is reduced by about 3.8 times and the conversion time was spent about 106 seconds.

• The Journal of the Acoustical Society of Korea
• /
• v.26 no.8
• /
• pp.409-414
• /
• 2007

In spite of many studies of the acoustic field visualization by using sonoluminescence phenomena, the visualization method has not been used widely because it needs high acoustic intensity to get the luminescence intensity enough to observe. Recently, the digital camera with high resolution and big memory makes it possible to get the digital image data even though the brightness of the image is too weak to observe with naked eyes. In this study we investigated the variation of sonoluminescence intensity with the acoustic intensity from an ultrasonic transducer. From this result, the inverse function, which makes the tendency of the variation to linear, was obtained. Using the order of the inverse function, we can expect a matching function. Applying the matching function to digital image data, the distribution of the histogram could be controlled appropriately and the image from relatively weak acoustic intensity could be enhanced by the method.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.9 no.1
• /
• pp.309-317
• /
• 2015

Recently many companies began to realize their visions for the sustainable growth with the advent of CSV(Creating Shared Values). Michael E. Porter, a Harvard Professor, claims that placing social value creation at the core of business strategy has the potential to uncover big opportunities for individual companies and that shared value can play a significant role in increasing competitive advantages while fostering social prosperity. In consequence, the various researches have illustrated how to get the opportunity for competitive advantages from building a social value proposition into corporate strategy, and considerable studies have been promoted heavily from the managerial perspective. However, due to the lack of capability converging information technology with business strategy, any research effort to identify technological or Internet-related issues and to link the issues to CSV does not exist. With Korean being a Internet leading country, the demands of researches analyzing core technology, information, and service utilizing Internet are rapidly growing. The study aims to find out Internet-related enablers for CSV. This paper describes the concepts and features of CSV, identifies emerging Internet-related issues toward the opportunity for competitive advantage, and then depicts the rigorous research endeavors in the areas of Internet information, technology, and services. As a result, 11 papers presented and selected as the outstanding papers at APIC-IST 2014 handle the issues to be brought together, which include: Wireless and Sensor Network, Image Processing and HCI, Big Data and Business Intelligence, Security & Privacy in Internet, SNS & Communication, Smart-Learning and e-Learning, and Internet Business Strategy. The study finally recommends indispensible terms for substantially vitalizing CSV.

• Journal of the Korea Society of Computer and Information
• /
• v.29 no.3
• /
• pp.11-20
• /
• 2024

Among the generative models in Artificial Intelligence (AI), especially Generative Adversarial Network (GAN) has been successful in various applications such as image processing, density estimation, and style transfer. While the GAN models including Conditional GAN (CGAN), CycleGAN, BigGAN, have been extended and improved, researchers face challenges in real-world applications in specific domains such as disaster simulation, healthcare, and urban planning due to data scarcity and unstable learning causing Image distortion. This paper proposes a new progressive learning methodology called Progressive Step Training (PST) based on the Auxiliary Classifier GAN (ACGAN) that discriminates class labels, leveraging the progressive learning approach of the Progressive Growing of GAN (PGGAN). The PST model achieves 70.82% faster stabilization, 51.3% lower standard deviation, stable convergence of loss values in the later high resolution stages, and a 94.6% faster loss reduction compared to conventional methods.

• The KIPS Transactions:PartA
• /
• v.10A no.6
• /
• pp.595-604
• /
• 2003

A Digital TV system is necessary for data Processing as well as video and audio processing. Especially in the case of interactive broadcasting, it should manage returning channel created by the Internet, PSTN, and so on. Because of many functionalities and multitasking jobs, it needs an Operating System. Embedded Linux as open source program can increase a cost effectiveness in market and has many advantages - reusable device drivers and application programs, more convenient developing environment using shell and file system, and easy problem resolution within Open Source Community. In this paper, we modified Embedded Linux kernel and cross developing environment for a big-endian system, redesigned devices for kernel execution, and configured system memory map in order to load a linux kernel. Also we developed an device driver for entire system control.