• 한국어정보학회:학술대회논문집
• /
• 한국어정보학회 2016년도 제28회 한글및한국어정보처리학술대회
• /
• pp.291-295
• /
• 2016

본 논문에서는 한국어 기본 유니트 단위로 WPM을 활용한 구글 플레이 스토어 앱의 댓글 감정분석을 수행하였다. 먼저 자동 띄어쓰기 시스템을 적용한 후, 어절단위, 형태소 분석기, WPM을 각각 적용하여 모델을 생성하고, 로지스틱 회귀(Logistic Regression), 소프트맥스 회귀(Softmax Regression), 서포트 벡터머신(Support Vector Machine, SVM)등의 알고리즘을 이용하여 댓글 감정(긍정과 부정)을 비교 분석하였다. 그 결과 어절단위, 형태소 분석기보다 WPM이 최대 25%의 향상된 결과를 얻었다. 또한 분류 과정에서 로지스틱회귀, 소프트맥스 회귀보다는 SVM 성능이 우수했으며, SVM의 기본 파라미터({'kernel':('linear'), 'c':[4]})보다 최적의 파라미터를 적용({'kernel': ('linear','rbf', 'sigmoid', 'poly'), 'C':[0.01, 0.1, 1.4.5]} 하였을 때, 최대 91%의 성능이 나타났다.

• 대한의용생체공학회:학술대회논문집
• /
• 대한의용생체공학회 1997년도 추계학술대회
• /
• pp.184-187
• /
• 1997

Small PACS based on a personal computer or CR has initially been designed to improve the performance of cost-effective PACS implementation. In that system, Wavelet compression scheme is newly adopted to store images hierarchically to storage unit, and retrieve and display images progressively or display workstation. In this compression method, image is decomposed into subclasses of image by wavelet transform, and then the subclasses of image are vector quantized using a multiresolution codebook.

• 한국정보과학회 언어공학연구회:학술대회논문집(한글 및 한국어 정보처리)
• /
• 한국정보과학회언어공학연구회 2016년도 제28회 한글 및 한국어 정보처리 학술대회
• /
• pp.291-295
• /
• 2016

본 논문에서는 한국어 기본 유니트 단위로 WPM을 활용한 구글 플레이 스토어 앱의 댓글 감정분석을 수행하였다. 먼저 자동 띄어쓰기 시스템을 적용한 후, 어절단위, 형태소 분석기, WPM을 각각 적용하여 모델을 생성하고, 로지스틱 회귀(Logistic Regression), 소프트맥스 회귀(Softmax Regression), 서포트 벡터머신(Support Vector Machine, SVM)등의 알고리즘을 이용하여 댓글 감정(긍정과 부정)을 비교 분석하였다. 그 결과 어절단위, 형태소 분석기보다 WPM이 최대 25%의 향상된 결과를 얻었다. 또한 분류 과정에서 로지스틱회귀, 소프트맥스 회귀보다는 SVM 성능이 우수했으며, SVM의 기본 파라미터({'kernel':('linear'), 'c':[4]})보다 최적의 파라미터를 적용({'kernel': ('linear','rbf', 'sigmoid', 'poly'), 'C':[0.01, 0.1, 1.4.5]} 하였을 때, 최대 91%의 성능이 나타났다.

• 대한전자공학회논문지SP
• /
• 제47권6호
• /
• pp.137-145
• /
• 2010

디지털 선박에서는 선박 내의 각종 센서로부터 측정된 디지털 데이터에 대한 정확하고 에너지 효율적인 관리가 필요하다. 그러나, 센서 네트워크에서 대용량 스트림 데이터를 제한된 네트워크, 전력, 프로세서를 이용하여 모든 센서 데이터를 전송하고 분석하는 것은 어렵고 효율적이지 못하다. 그러므로, 연속적으로 입력되는 데이터를 사전에 분류하여 특성에 따라 선택적으로 데이터를 처리하는 데이터 분류 기법이 요구된다. 본 논문에서는 디지털 선박 내에 다수 개의 센서(온도, 습도, 조도, 음성 센서)를 배치하고 효율적인 입력 스트림 처리를 위해서 슬라이딩 윈도우 기반으로 다중 Support Vector Machine(SVM) 알고리즘을 이용하여 사전 분류(pre-clustering)한 후 요약된 정보를 해쉬 테이블로 관리하는 효율적인 처리 기법을 제안한다. 해쉬테이블을 이용하여 다차원 스트림 데이터의 저장될 레코드 순서를 빠르게 찾아 저장 및 검색함으로서 처리 속도가 향상되고 메모리에 해쉬 테이블 만을 유지하면 되므로 메모리 사용량이 감소한다. 35,912개의 데이터 집합을 사용하여 실험한 결과 제안 기법의 정확도와 처리 성능이 향상되었다.

• 한국정보처리학회논문지
• /
• 제6권4호
• /
• pp.1098-1105
• /
• 1999

분산 공유 메모리(distributed shared memory) 시스템에서 캐쉬는 메모리 접근 지연과 통신 부하 줄임으로 성능을 향상시킬 수 있으나 캐쉬일관성 문제를 해결하여야 한다. 본 논문은 DSM 시스템에서 캐쉬일관성 문제를 해결하고 성능을 향상시킬 수 있는 새 디렉터리 프로토콜을 제안한다. 캐시 일관성을 유지하기 일정거리 이내에 있는 처리기는 전체 디렉터리 기법처럼 비트 벡터를 사용하여 통신 오버헤드를 줄일 수 있다. 그리고 일정거리 이상에 있는 처리기는 포인터를 디렉터리 풀에 저장한다. 이 비트 벡터와 디렉터리 풀의 사용은 불필요한 캐쉬 무효화를 방지하므로 시스템의 성능을 향상시킬 수 있다. 제안한 기법은 제한 디렉터리 기법보다 통행량을 66%까지 줄일 수 있으며 동적할당 디렉터리 기법보다 디렉터리 접근 회수도 27%까지 각각 줄일 수 있다.

• Journal of Biosystems Engineering
• /
• 제33권6호
• /
• pp.416-422
• /
• 2008

This study was performed to design and construct a digital soil shear stress sensor in order to replace the conventional devices for measuring soil shear property. The developed digital shear stress measuring device can store measured data with GPS position information as a vector format into a computer. Based on the experiments at various field conditions, the measuring characteristic of the device was quite similar to that of a conventional device, SR-2 that has been a major tool to measure the soil shear property. It was concluded that the digital shear stress measuring device was an effective and comprehensive sensor for measuring soil shear property.

• 대한전기학회:학술대회논문집
• /
• 대한전기학회 2005년도 학술대회 논문집 정보 및 제어부문
• /
• pp.261-263
• /
• 2005

In this paper, we implement an identification system using the vein image of the hand. The vein pattern is obtained in the grey-scale 2D image through the infrared-red imaging from back of the hand. Since the frame has lack of clearance, we use some enhancing methods such as the complement, addition, and multiplication to the image to increase the contrast. After Wiener filtering for smoothness of the vein pattern, we transform the image into the binary image with mean function. The binarized image is session thinned and the cross-points in the vein tree are obtained by calculating the number of pixels connected because the image is shaped as a tree. We choose the point and find the nearest to the center if it has majority, where we find the two end points of the selected line. We can get the angle between the two lines joined at the cross-point and store its coordinates, angle, and label the values. The values are used as the feature vector of the vein pattern. This procedure is similar to the human cognition sequences. It is shown that the proposed method is simple for the vein recognition.

• 대한가정학회지
• /
• 제39권10호
• /
• pp.55-67
• /
• 2001

The purpose of this study were 1) to find out the benefits sought factors and segment the customers of internet shopping mall, 2) to find out the store image factors of internet shopping mall, and 3) to analyze the internet shopping mall market using perceptual map of segmented groups. The questionnaires dealing with attribute dimension of internet shopping mall image, benefits sought, and demographic variables were selected from the previous studies or were developed for this study. The data from 319 respondents which were collected through the internet survey site was analyzed by factor analysis, cluster analysis, one-way ANOVA, and $X^2$-test. The results of this study were as follows: 1. Benefit sought by consumer in internet shopping malls was found to include six different factors: assortments of products, search efficiency, brand/fashionability, delivery convenience, promotion service and informativeness. 2. As a result of subdividing the consumers, four distinctive groups were formed on the basis of benefit factors: multi-benefit oriented group, convenience oriented group, brand oriented group and low-benefit oriented group. Demographic traits such as education and income level were proven to significantly differentiate the benefit segments. 3. In the structural components of internet shopping-malls image, product/information, service/convenience and economy were drawn from attribute dimensions. 4. 12 perceptual maps of internet shopping mall image were constructed and each ideal vector were drawn.

• Journal of Communications and Networks
• /
• 제18권2호
• /
• pp.261-269
• /
• 2016

In this paper, the impact of energy constraints on a two-hop network with a source, a relay and a destination under random medium access is studied. A collision channel with erasures is considered, and the source and the relay nodes have energy harvesting capabilities and an unlimited battery to store the harvested energy. Additionally, the source and the relay node have external traffic arrivals and the relay forwards a fraction of the source node's traffic to the destination; the cooperation is performed at the network level. An inner and an outer bound of the stability region for a given transmission probability vector are obtained. Then, the closure of the inner and the outer bound is obtained separately and they turn out to be identical. This work is not only a step in connecting information theory and networking, by studying the maximum stable throughput region metric but also it taps the relatively unexplored and important domain of energy harvesting and assesses the effect of that on this important measure.

• Journal of Information Processing Systems
• /
• 제17권3호
• /
• pp.599-614
• /
• 2021

This paper proposes a digital signage system based on an intelligent recommendation model. The proposed system consists of a server and an edge. The server manages the data, learns the advertisement recommendation model, and uses the trained advertisement recommendation model to determine the advertisements to be promoted in real time. The advertisement recommendation model provides predictions for various products and probabilities. The purchase index between the product and weather data was extracted and reflected using correlation analysis to improve the accuracy of predicting the probability of purchasing a product. First, the user information and product information are input to a deep neural network as a vector through an embedding process. With this information, the product candidate group generation model reduces the product candidates that can be purchased by a certain user. The advertisement recommendation model uses a wide and deep recommendation model to derive the recommendation list by predicting the probability of purchase for the selected products. Finally, the most suitable advertisements are selected using the predicted probability of purchase for all the users within the advertisement range. The proposed system does not communicate with the server. Therefore, it determines the advertisements using a model trained at the edge. It can also be applied to digital signage that requires immediate response from several users.