• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.13 no.4
• /
• pp.1989-2011
• /
• 2019

Vehicle detection based on aerial images is an interesting and challenging research topic. Most of the traditional vehicle detection methods are based on the sliding window search algorithm, but these methods are not sufficient for the extraction of object features, and accompanied with heavy computational costs. Recent studies have shown that convolutional neural network algorithm has made a significant progress in computer vision, especially Faster R-CNN. However, this algorithm mainly detects objects in natural scenes, it is not suitable for detecting small object in aerial view. In this paper, an accurate and effective vehicle detection algorithm based on Faster R-CNN is proposed. Our method fuse a hyperactive feature map network with Eltwise model and Concat model, which is more conducive to the extraction of small object features. Moreover, setting suitable anchor boxes based on the size of the object is used in our model, which also effectively improves the performance of the detection. We evaluate the detection performance of our method on the Munich dataset and our collected dataset, with improvements in accuracy and effectivity compared with other methods. Our model achieves 82.2% in recall rate and 90.2% accuracy rate on Munich dataset, which has increased by 2.5 and 1.3 percentage points respectively over the state-of-the-art methods.

• Journal of Digital Convergence
• /
• v.17 no.11
• /
• pp.233-240
• /
• 2019

In order to predict and respond to cyber attacks, a number of security companies quickly identify the methods, types and characteristics of attack techniques and are publishing Security Intelligence Reports(SIRs) on them. However, the SIRs distributed by each company are huge and unstructured. In this paper, we propose a framework that uses five analytic techniques to formulate a report and extract key information in order to reduce the time required to extract information on large unstructured SIRs efficiently. Since the SIRs data do not have the correct answer label, we propose four analysis techniques, Keyword Extraction, Topic Modeling, Summarization, and Document Similarity, through Unsupervised Learning. Finally, has built the data to extract threat information from SIRs, analysis applies to the Named Entity Recognition (NER) technology to recognize the words belonging to the IP, Domain/URL, Hash, Malware and determine if the word belongs to which type We propose a framework that applies a total of five analysis techniques, including technology.

• Journal of Families and Better Life
• /
• v.15 no.2
• /
• pp.71-71
• /
• 1997

This study aims ai the extraction of what is universally to be taught in modern Korean Families and its scalization. That is, it attempts to provide the criterion by which we can determine what to teach adolescent in the family, not in society or school. For this purpose, this study firstly reviews the preceeding studies on the subject. As a result of the review, we postulate the hypothetical structure consisting of 11 domains. Secondary, we gather the parent's view on the topic by interviewing 496 parents with teenagers. On the basis of this study, we can construct the questionnaire(Likert scale; 5 point). After we conduct an extensive empirical research(346 parents) in order to generalize 195 items of the workedout questionnaire. We apply factor analysis(principal axis factoring, oblique (promax) rotation) in the verification of the validity. As the consequence, we select 66 items consisting, 10 factors, which explain 68% of common variance. We name the 10 educational factors extracted in the scale as follows ; The Sense of Value, Communal Society, Sex, Esteem for an ancestor & a traditional way of life, Parent-Child Relationship, the Culture life within the family, The Guide of Learning Way, Setting up the way of life, The control of one's life, Friendship. The reliability of the scale is the cronbach =0.91 which turns out to be satisfactory.

• Journal of Korea Multimedia Society
• /
• v.7 no.10
• /
• pp.1354-1369
• /
• 2004

This paper presents a logical structure analysis method which transforms Web documents into XML ones. The proposed method consists of three phases: visual grouping, element identification, and logical grouping. To produce a logical structure more accurately, the proposed method defines a document model that is able to describe logical structure information of topic-specific document class. Since the proposed method is based on a visual structure from the visual grouping phase as well as a document model that describes logical structure information of a document type, it supports sophisticated structure analysis. Experimental results with HTML documents from the Web show that the method has performed logical structure analysis successfully compared with previous works. Particularly, the method generates XML documents as the result of structure analysis, so that it enhances the reusability of documents.

• Annual Conference of KIPS
• /
• 2017.04a
• /
• pp.873-876
• /
• 2017

최근 정보검색의 효율성을 위해 데이터를 분석하여 해당 데이터를 가장 잘 나타내는 연관단어를 추출 및 추천하는 연구가 활발히 이루어지고 있다. 현재 관련 연구들은 출현 빈도수를 사용하는 방법이나 LDA와 같은 기계학습 기법을 활용해 데이터를 분석하여 연관단어를 생성하는 방법을 제안하고 있다. 기계학습 기법은 결과 값을 찾는데 사용되는 특징들을 전문가가 직접 설계해야 하며 좋은 결과를 내는 적절한 특징을 찾을 때까지 많은 시간이 필요하다. 또한, 파라미터들을 직접 설정해야 하므로 많은 시간과 노력을 필요로 한다는 단점을 지닌다. 이러한 기계학습 기법의 단점을 극복하기 위해 인공신경망을 다층구조로 배치하여 데이터를 분석하는 딥러닝이 최근 각광받고 있다. 본 논문에서는 기존 기계학습 기법을 사용하는 연관단어 추출연구의 한계점을 극복하기 위해 딥러닝을 활용한다. 먼저, 인공신경망 기반 단어 벡터 생성기인 Word2Vec를 사용하여 다양한 텍스트 데이터들을 학습하고 룩업 테이블을 생성한다. 그 후, 생성된 룩업 테이블을 바탕으로 인공신경망의 한 종류인 합성곱 신경망을 활용하여 사용자가 입력한 주제어와 관련된 최근 뉴스데이터를 분석한 후, 주제별 최신 연관단어를 추출하는 시스템을 제안한다. 또한 제안한 시스템을 통해 생성된 연관단어의 정확률을 측정하여 성능을 평가하였다.

• Journal of Fisheries and Marine Sciences Education
• /
• v.28 no.3
• /
• pp.872-880
• /
• 2016

The purpose of this study is to find the research trends relating to flipped learning through keyword network analysis. For investigating this topic, final 100 papers (removed due to overlap in all 205 papers) were selected as subjects from the result of research databases such as RISS, DBPIA, and KISS. After keyword extraction, coding, and data cleaning, we made a 2-mode network with final 202 keywords. In order to find out the research trends, frequency analysis, social network structural property analysis based on co-keyword network modeling, and social network centrality analysis were used. Followings were the results of the research: (a) Achievement, writing, blended learning, teaching and learning model, learner centered education, cooperative leaning, and learning motivation, and self-regulated learning were found to be the most common keywords except flipped learning. (b) Density was .088, and geodesic distance was 3.150 based on keyword network type 2. (c) Teaching and learning model, blended learning, and satisfaction were centrally located and closed related to other keywords. Satisfaction, teaching and learning model blended learning, motivation, writing, communication, and achievement were playing an intermediary role among other keywords.

• Journal of Yeungnam Medical Science
• /
• v.21 no.1
• /
• pp.12-22
• /
• 2004

The management of coincident glaucoma and cataract is not only a common clinical challenge but also an important research topic in the ophthalmic surgical field. The purpose of this article is to compare the different surgical options on the basis of their achievable postoperative intraocular pressure (IOP) control, success rates, and complication rates reported in the related literature, and to give advice on how to manage typical situations of patients with both glaucoma and cataract. Main topics were focused on indications and rationale of 3 surgical options (only cataract surgery first and later trabeculectomy, only trabeculectomy first and later cataract surgery, or simultaneous combined surgery). Modern clear corneal cataract extraction techniques resulted in a modest intermediate-term reduction of IOP and has considerably improved the success rates of combined glaucoma and cataract surgery. It also enabled future trabeculectomy to be successfully performed at a later date if necessary. Trabeculectomy alone achieved better IOP regulation than phacotrabeculectomy (combined surgery), but subsequent cataract surgery may compromise preexisting filtering bleb. Combined surgery augmented with mitomycin C achieved a lower IOP than combined surgery alone but had a higher complication rate. In conclusion, the choice of the preferred surgical method should be determined according to the target pressure, the amount of glaucomatous damage, and the grade of visual disturbance caused by the cataract. Phacotrabeculectomy with adjunctive mitomycin C offers visual improvement and achieves the best IOP lowering of all types of combined glaucoma and cataract surgery currently used but is associated with potentially sight-threatening complications.

• The Journal of the Korea Contents Association
• /
• v.10 no.5
• /
• pp.29-36
• /
• 2010

Past clustering researches are focused on extraction of keyword for word similarity grouping. However, too many candidates to compare and compute bring high complexity, low speed and low accuracy. To overcome these weaknesses, this paper proposed a topical web document clustering model using not only keyword but also named entities such as person name, organization, location, and so on. By several experiments, we prove effects of our model compared with traditional model based on only keyword and analyze how different effects show according to characteristics of document collection.

• Journal of KIISE:Computing Practices and Letters
• /
• v.14 no.3
• /
• pp.336-340
• /
• 2008

Sentiment classification is a recent subdiscipline of text classification, which is concerned not with the topic but with opinion. In this paper, we present a Korean sentence and document classification system using effective sentiment features. Korean sentiment classification starts from constructing effective sentiment feature sets for positive and negative. The synonym information of a English word thesaurus is used to extract effective sentiment features and then the extracted English sentiment features are translated in Korean features by English-Korean dictionary. A sentence or a document is represented by using the extracted sentiment features and is classified and evaluated by SVM(Support Vector Machine).

• Journal of Families and Better Life
• /
• v.14 no.1
• /
• pp.65-84
• /
• 1996

This study aims at the extraction of what is universally to be taught in modern Korean Families and its scalization. That is it attempts to provide the criterion by which we can determine what to teach children in the family not in society or school. For this purpose this study firstly reviews the preceeding studies on the subject. As a result of the review we postulate the hypothetical structure consisting of 12 domains. Secdndly we gather the parent's view on the topic by interviewing 192 parents of the students. On the basis of this study we can construct the questionnaire(Likert scale; 5 point) After we conduct an extensive empirical research(518 parents) in order to generalize 179 items of the worked-out questionnaire. We apply factor analsis(principal axis factoring oblique (promax) rotation) in the verfication of the validity. As the consequence we select 87 items consisting 15 factors which explain 71% of common variance. We name the 15 educational factors extracted in the scale as follows; Values & Sociality Training for basic living habits The maintenance of the culture life on the family Religious life Learning guide Economic life Patriotism Independent living Table manners The management of commodity The preservation of family unity Sex & a Frendship with the opposite sex Esteem for a traditional way of life Respect for the Old Filial piety. The reliabiliy of the scale is the cronbach a=.96 which turns out to be satisfactory.