• Proceedings of the Korea Information Processing Society Conference
• /
• 2013.05a
• /
• pp.281-284
• /
• 2013

한국전자통신연구원(ETRI)에서는 2010 년부터 2015 년까지 5 년간에 걸쳐 모바일 플랫폼 기반 대화모델이 적용된 자연어 음성인터페이스 기술을 개발하고 있다. 2010 년에는 대화 시스템의 전반적인 모습을 설계하였고, 2011 년에는 대상 도메인으로 도시 관광용 영어 대화 시스템을, 2012 년에는 대상 도메인으로 차량공조, 응급조치 등과 같은 차량 정보 서비스용 한국어 대화 시스템을 개발하였다. 본 논문에서는 2012 년에 개발한 차량 정보 서비스용 한국어 대화 시스템을 기술하는 것을 목표로 한다. 차량 정보 서비스용 한국어 대화 시스템의 성능 평가는 운전 경험이 있는 평가자 20 명에 의해 이루어졌다. 평가자는 웹 평가 도구에 원격으로 접속하여 주어진 40 개의 차량 정보 관련 대화 미션을 태스크로 하여 차량 정보 서비스용 대화 시스템과 대화를 하였다. 평가는 태스크 성공률과 대화턴 성공률로 나누어 측정되었으며 태스크 성공률은 87.8%, 대화턴 성공률은 86.7%였다.

• Food Science and Industry
• /
• v.51 no.3
• /
• pp.229-237
• /
• 2018

The objective of this paper is to understand the history, current status, and future of Korea meat processing industry. The numbers of businesses, employees and distribution in the domestic meat processing industry have steadily increased year by year. The trends of the Korea meat processing market are being launched with customized products due to rapid changes in consumer's lifestyle. Futhermore, the misconceptions against meat products have been still going on these days. Thus, meat products are perceived as a representative food for fast food and junk food to consumers, and growth of meat processing industry was slow-down due to concerns that some additives contained in meat products, such as phosphate and nitrite may not be good for human health, as World Health Organization (WHO) reported a couple of years ago. Therefore, future meat processing industries should be developed safe, and high quality consumer-oriented products. Finally, it will be able to achieve the unlimited development of the Korea meat processing industry by monitoring rapidly changing consumer needs, improving awareness and producing high quality meat products.

• International Journal of Advanced Culture Technology
• /
• v.10 no.2
• /
• pp.252-259
• /
• 2022

Artificial intelligence is used in fusion with image processing techniques using cameras. Image processing technology is a technology that processes objects in an image received from a camera in real time, and is used in various fields such as security monitoring and medical image analysis. If such image processing reduces the accuracy of recognition, providing incorrect information to medical image analysis, security monitoring, etc. may cause serious problems. Therefore, this paper uses a mixture of YOLOv4-tiny model and image processing algorithm and uses the COCO dataset for learning. The image processing algorithm performs five image processing methods such as normalization, Gaussian distribution, Otsu algorithm, equalization, and gradient operation. For RGB images, three image processing methods are performed: equalization, Gaussian blur, and gamma correction proceed. Among the nine algorithms applied in this paper, the Equalization and Gaussian Blur model showed the highest object detection accuracy of 96%, and the gamma correction (RGB environment) model showed the highest object detection rate of 89% outdoors (daytime). The image binarization model showed the highest object detection rate at 89% outdoors (night).

• ETRI Journal
• /
• v.45 no.2
• /
• pp.318-328
• /
• 2023

Recently, embedded systems, such as mobile platforms, have multiple processing units that can operate in parallel, such as centralized processing units (CPUs) and neural processing units (NPUs). We can use deep-learning compilers to generate machine code optimized for these embedded systems from a deep neural network (DNN). However, the deep-learning compilers proposed so far generate codes that sequentially execute DNN operators on a single processing unit or parallel codes for graphic processing units (GPUs). In this study, we propose PartitionTuner, an operator scheduler for deep-learning compilers that supports multiple heterogeneous PUs including CPUs and NPUs. PartitionTuner can generate an operator-scheduling plan that uses all available PUs simultaneously to minimize overall DNN inference time. Operator scheduling is based on the analysis of DNN architecture and the performance profiles of individual and group operators measured on heterogeneous processing units. By the experiments for seven DNNs, PartitionTuner generates scheduling plans that perform 5.03% better than a static type-based operator-scheduling technique for SqueezeNet. In addition, PartitionTuner outperforms recent profiling-based operator-scheduling techniques for ResNet50, ResNet18, and SqueezeNet by 7.18%, 5.36%, and 2.73%, respectively.

• Annual Conference on Human and Language Technology
• /
• 2019.10a
• /
• pp.533-536
• /
• 2019

의존 구문 분석은 문장 구성 성분 간의 의존 관계를 분석하는 태스크로, 자연어 이해의 대표적인 과제 중 하나이다. 본 논문에서는 한국어 의존 구문 분석의 성능 향상을 위해 Deep Bi-Affine Network와 Left to Right Dependency Parser를 적용하고, 새롭게 한국어의 언어적 특징을 반영한 Right to Left Dependency Parser 모델을 제안한다. 3개의 의존 구문 분석 모델에 단어 표현을 생성하는 방법으로 ELMo, BERT 임베딩 방법을 적용하고 여러 종류의 모델을 앙상블하여 세종 의존 구문 분석 데이터에 대해 UAS 94.50, LAS 92.46 성능을 얻을 수 있었다.

• Journal of Distribution Science
• /
• v.22 no.6
• /
• pp.57-68
• /
• 2024

Purpose: Dynamic capabilities are acknowledged as pivotal factors facilitating the generation of competitive advantage and performance for enterprises across diverse sectors, notably within the food processing industry. This study endeavors to explore and examine the direct and indirect impact of dynamic capability factors, encompassing innovation capability, market orientation (customer orientation, supply chain orientation, competitor orientation), and digital orientation on the business performance of Vietnamese food processing enterprises. Research design, data and methodology: Research data was collected from 239 food processing enterprise managers and analyzed using partial least square - structural equation modeling. Results: With the exception of digital orientation, the study found evidence of the role of innovation capability and market orientation (customer orientation, supply chain orientation, competitor orientation) in the overall competitive advantage of Vietnamese food processing enterprises. Furthermore, the study underscores the contributions of innovation capability, market orientation and digital orientation in fostering performance of Vietnamese food processing enterprises. It also identifies the mediating role of competitive advantage in the relationship between dynamic capabilities and performance. Conclusions: Consequently, the study proposes a number of implications for food processing enterprise managers with the aim of establishing and enhancing competitive advantage and business performance in the future.

• Journal of KIISE:Software and Applications
• /
• v.26 no.8
• /
• pp.946-954
• /
• 1999

웹의 등장으로 전자 업무 처리는 급격한 성장을 보여왔다. 많은 업계에서는 업무 처리에 있어 핵심이 되는 트랜잭션을 웹에서 처리하기 위한 시스템 개발에 많은 노력을 기울여왔다. 그러나 대부분의 웹 트랜잭션 연구는 주로 CGI 방식의 트랜잭션 처리 성능의 한계를 극복하거나, 여러 HTTP 요청간의 상태 유지 방법 또는 시스템의 확장성 등에 대해 연구가 집중된 반면, 부적합한 웹 환경으로 유발되는 시스템의 일관성 손실에 대해서는 많은 관심을 받지 못했다.본 논문에서는 웹 상에서 신뢰성 있는 트랜잭션을 처리하는 경우 발생되는 세 가지 문제점을 지적하고, 이를 막기 위해 웹 트랜잭션 처리 시스템인 WebTP에서 제공하는 해결책을 제안한다. 제안한 해결책은 쿠키 로깅, 웹 페이지 로깅, 그리고 일방적으로 철회된 트랜잭션의 상태를 유지하는 기법을 근간으로 한다. 또한 제안된 방법은 유사 구조의 타 웹 트랜잭션 처리 시스템에서도 적용 가능하다.Abstract Electronic business processing has drastically grown since the advent of the World Wide Web. As transaction processing is at heart of business processing, many companies have been interested in transaction processing on the Web. However, little attention has been paid to supporting consistent transaction processing on the Web, while most of the work has been focused on performance enhancement, processing multiple HTTP requests in a single context, and scalability.This paper identifies three problems in consistent transaction processing on the Web, and proposes corresponding solutions for each problem. The results are implemented in the Web transaction processing system WebTP. These solutions are based on cookie logging, Web page logging and keeping information for unilaterally aborted transactions. The solutions proposed in this paper also can be applied to other Web transaction processing system of similar architecture.

• Aerospace Engineering and Technology
• /
• v.13 no.1
• /
• pp.174-183
• /
• 2014

The Value-Added Processing System(VAPS) was developed for post-processing the KOMPSAT imagery. Recently software version and hardware specification of VAPS were changed for improving the VAPS performance. The purpose of this study is to describe about the improvement of existing VAPS(ver.1.0) and systematically evaluate the performance of the improved VAPS(ver.2.0). To this end, test-bed areas in South and North Korea were selected and then image processing tests were conducted using KOMPSAT-2 and KOMPSAT-3 imagery in both areas. In conclusion, VAPS(ver.2.0) had an ability to generate the high level products like ortho images and mosaic images. Image processing time using the Graphic Processing Unit(GPU) on ver.2.0 was enhanced up to 10 times than ver.1.0.

• The Mathematical Education
• /
• v.59 no.4
• /
• pp.389-403
• /
• 2020

This study aims to examine how information processing competency as one of the mathematical competencies has been interpreted and applied in mathematics education by analyzing tasks in middle school mathematics textbooks under the 2015 revised national curriculum. Based on the sub-elements of information processing competency organized by Park et al.(2015), we analyzed 191 tasks in 30 different middle school mathematics textbooks using descriptive statistics and ANOVA. Also, we investigated the meaning of information processing competency embedded in the tasks by distinguishing the characteristics of several different types of tasks. The results from this study showed that the number of tasks related to information processing competency in mathematics textbooks was too small and there was a huge difference across the textbooks in terms of the sub-elements. Even though there were four sub-elements of information processing competency, 'the use of manipulative and technological tools' was extremely dominant in the tasks in general. Even many of them used technology and manipulatives superficially. Furthermore, any textbook did not provide tasks dealing with all the four sub-elements. Such an unbalanced and fragmented approach to information processing competency could produce biased knowledge and insufficient experiences for information processing competency. It calls for further investigation and discussion about how to improve information processing competency in school mathematics.

• Journal of KIISE
• /
• v.41 no.11
• /
• pp.958-966
• /
• 2014

Wireless sensor networks offer a distributed processing environment. Many sensor nodes are deployed in fields that have limited resources such as computing power, network bandwidth, and electric power. The sensor nodes construct their own networks automatically, and the collected data are sent to the sink node. In these traditional wireless sensor networks, network congestion due to packet flooding through the networks shortens the network life time. Clustering or in-network technologies help reduce packet flooding in the networks. Many studies have been focused on saving energy in the sensor nodes because the limited available power leads to an important problem of extending the operation of sensor networks as long as possible. However, we focus on the execution time because clustering and local distributed processing already contribute to saving energy by local decision-making. In this paper, we present a cooperative processing model based on the processing timeline. Our processing model includes validation of the processing, prediction of the total execution time, and determination of the optimal number of processing nodes for distributed processing in wireless sensor networks. The experiments demonstrate the accuracy of the proposed model, and a case study shows that our model can be used for the distributed application.