• Science of Emotion and Sensibility
• /
• v.11 no.4
• /
• pp.501-510
• /
• 2008

The purpose of this study was to investigate whether the orientation of the head position across different categories affect reaction time and accuracy of object recognition. Fifty four right handed undergraduate students were participated in the experiment. Participants performed the word-picture matching tasks, which were different in terms of head direction of object (i.e., Left-headed or Right-headed) and object category (i.e., natural : animal or artificial : tool). Participants were asked to decide whether each picture matched the word which was followed by the picture. For accuracy, no statistically significant difference was found for both animal and tool pictures due to the ceiling effect. Interaction effect of category and orientation were statistically significant, whereas only the main effect of category was significant. In the animal condition, faster reaction times were observed for left to right than right to left presentation, while no statistical significant difference was found in the tool condition. The orientation of the object's canonical representation was different across different categories. The faster RT for the animal condition implies that the canonical representation for animal is left-headed. This could be due to the orientation of the face.

• Korean Journal of Cognitive Science
• /
• v.19 no.2
• /
• pp.177-203
• /
• 2008

The present study was aimed to compare the semantic systems represented by the lexicon between L1 and L2 in late Korean-English bilinguals. The participants performed the word-picture matching task. the task was to decide whether the pictures represent the previously presented words' meaning. The words were the basic level categories. The stimuli were consisted of common object belonged to two different semantic categories (natural and artificial). To control the translation strategies, the SOA were manipulated as 650ms(Exp. 1) and 250ms(Exp. 2). No translation effort was found in the comparison of the two experiments. In both experiment, the RTs were faster in L1 rendition, and it took longer to decide the stimuli in natural categories than with artificial ones in L1. However, this category effect was not observed in L2. The results showed the differences in the organization of semantic representations in the brain through the bilinguals' two languages. While L1 semantic knowledge might be more systematically organized, that of L2 seems to be less well organized, at least by late bilinguals who participated in the present study.

• Journal of Korean Library and Information Science Society
• /
• v.48 no.4
• /
• pp.21-47
• /
• 2017

This study aims to analyze features of topics about AI(Artificial Intelligence) which is gaining a massive attention these days. Newspaper articles published from 2016 to June, 2017 were selected to analyze key subjects. The reason why the period was selected is people started to get attention on AI since 2016 as AlphaGo came out and gave a shock. The number of coded main message was 1,210 in 525 newspaper articles in total. The messages were categorized as three subject categories: the seven major categories, 62 middle categories. and minor categories. The seven major categories contains issues such as AI research, AI application, AI business, AI era, AI argument, AlphaGo, and other topics. The first features of issues about AI found in the major subject categories is that they are various and complicate. Second, it is important that social and policy-level issues related AI, such as job losses, misuse, and error should be dealt with to utilize AI safely. Last, issues related the role of human and revolution of education system in the AI era were shown as subjects which are important but hard to discuss.

• Annual Conference on Human and Language Technology
• /
• 2022.10a
• /
• pp.251-256
• /
• 2022

범주 불균형은 분류 모델이 다수 범주에 편향되게 학습되어 소수 범주에 대한 분류 성능을 떨어뜨리는 문제를 야기한다. 언더 샘플링 기법은 다수 범주 데이터의 수를 줄여 소수 범주와 균형을 이루게하는 대표적인 불균형 해결 방법으로, 텍스트 도메인에서의 기존 언더 샘플링 연구에서는 단어 임베딩과 랜덤 샘플링과 같은 비교적 간단한 기법만이 적용되었다. 본 논문에서는 트랜스포머 기반 문장 임베딩과 군집화 기반 샘플링 방법을 통해 텍스트 데이터의 정보 손실을 최소화하는 언더샘플링 방법을 제안한다. 제안 방법의 검증을 위해, 감성 분석 실험에서 제안 방법과 랜덤 샘플링으로 추출한 훈련 세트로 모델을 학습하고 성능을 비교 평가하였다. 제안 방법을 활용한 모델이 랜덤 샘플링을 활용한 모델에 비해 적게는 0.2%, 많게는 2.0% 높은 분류 정확도를 보였고, 이를 통해 제안하는 군집화 기반 언더 샘플링 기법의 효과를 확인하였다.

• 한국정보교육학회:학술대회논문집
• /
• 2021.08a
• /
• pp.167-173
• /
• 2021

The purpose of this study is to discuss the types of algorithms and data categories in AI education for elementary school students. The study surveyed 11 pre-elementary teachers after providing education and practice on various data, artificial intelligence algorithm, and AI education platform for 15 weeks. The categories of data and algorithms considering the elementary school level, and educational tools were presented, and their suitability was analyzed. Through the questionnaire, it was concluded that it is most suitable for the teacher to select and preprocess data in advance according to the purpose of the class, and the classification and prediction algorithms are suitable for elementary AI education. In addition, it was confirmed that Entry is most suitable as an AI educational tool, and materials that explain mathematical knowledge are needed to educate the concept of learning of AI. This study is meaningful in that it specifically presents the categories of algorithms and data with in AI education for elementary school students, and analyzes the need for related mathematics education and appropriate AI educational tools.

• Proceedings of the Korean Operations and Management Science Society Conference
• /
• 2006.05a
• /
• pp.291-298
• /
• 2006

패턴인식 문제에서 한 범주에 속한 데이터의 수가 다른 범주에 속한 데이터의 수보다 극히 많거나 적으면 데이터 불균형이 발생했다고 한다. Support Vector Machine(SVM)은 다른 기계 학습 알고리즘들과 마찬가지로 학습에 사용되는 데이터의 범주간 비율이 거의 비슷하다는 가정 하에서 학습을 하고 예측 결과를 도출하게 된다. 그러나 실제 문제에서는 데이터의 불균형이 발생하는 경우가 매우 빈번하며, 이러한 경우에는 모델의 성능이 매우 저하되는 문제점이 발생한다. 본 논문에서는 실제로 데이터 불균형이 SVM의 분류 결과에 어떠한 영향을 미치는지를 2차원 인공 데이터를 통하여 알아본다. 그리고 이러한 데이터 불균형을 해소하기 위하여 Under-Sampling 기반 앙상블 SVM을 제안하였다. 제안된 방법을 두 가지 인공 데이터에 적용하여 본 결과, 제안된 방법은 데이터 불균형을 해소하기 위해 사용되는 기존의 방법들에 비하여 소수 범주에 속하는 데이터의 수가 매우 적고 데이터의 불균형이 매우 심한 경우에도 높은 성능과 안정성을 갖는 효과적인 방법이라는 것이 입증되었다.

• Proceedings of the Korean Information Science Society Conference
• /
• 2001.04b
• /
• pp.508-510
• /
• 2001

3차원 물체의 인식과 공간 정보를 추출해 내는 것이 물체인식의 주요 목적이다. 본 논문에서는 평면의 표면을 갖는 기하학적 물체들을 인식하는데 인공신경망이 적용 가능함이 조사되었다. 물체인식을 위한 모델들은 CAD모델들로부터 자동적으로 추출되며, 획득된 물체의 영상과 일치하는 물체의 국면(aspect)과의 매칭은 조건만족 인경신경망을 이용하여 매칭-오차를 최소화시키는 방법을 처리되었다. 인식된 물체의 국면이 어느 방향에서 획득되었는지에 대한 정보(Aspect's view direction)는 검색된 가시 평면들의 분포로부터 추출됨을 ART와 같은 인공신경망을 이용하여 실시간으로 복원할 수 있음을 보였다. 대표적이 측정방향과 이 측정방향으로부터의 편차들을 한 범주에 넣고 학습을 통해 정확한 측정방향 정보들을 구하며, 획득된 3차원 물체의 영상들에 따라 자동적으로 측정방향범주 들이 추가되도록 한다.

• Korean Journal of Cognitive Science
• /
• v.29 no.1
• /
• pp.1-16
• /
• 2018

In this study, we investigated age differences of preference and trust ratings when the appearance of an artificial intelligent speaker resembles a human face. The appearance of the artificial intelligent speaker was presented in seven levels from robot face to human face. In addition, face stimuli were divided into gender (male and female) and age (20s / 60s). Participants evaluated the reliability and likability of each face stimulus on a 7-point scale. The results show that younger adults tend to prefer the face that was halfway between the robot and the human face, while older adults evaluated that the perceived reliability and likability were higher when the stimuli resembled the human face. When asked to choose the most preferred of the four face categories, all participants chose a younger face. However, with additional conditions including emoticon face and empty condition, older adults still preferred human face, while younger adults preferred emoticon face and empty condition. Taken together, older adults are more receptive to human faces than robotic faces in the context of artificial intelligence speakers. Because artificial intelligent speakers can play an important role in the elderly living alone, the present study will be a good reference in the design and development of artificial intelligent speakers for the elderly users.

• Korean Journal of Cognitive Science
• /
• v.21 no.4
• /
• pp.505-534
• /
• 2010

The purpose of this study was aimed to compare the lexical access and representation of semantic system in the bilinguals. The participants(late Korean-English bilinguals) performed the word-picture matching task. The task was to decide whether the pictures presented after the words(basic-level categories) represent the Korean(L1) or English(L2) words' meaning or not. The stimuli were consisted of common object belonged to four different categories(animal, part of body, clothes, tool). To control the translation strategies, the SOA(stimulus onset asynchrony) were manipulated as 650ms(Exp. 1) and 200ms(Exp. 2). In both experiment, the RTs were faster in L1 condition. The decision time of the part of body categories were shorter than the animal in L1 condition. In L2 condition, clothes were responded faster than the tools. The differences of the lexical access time implied that the bilingual semantic system seemed to be structured by more sub-level categories than the super-level, living or non-living things, and the ways to access the bilingual lexicon might be differentiated according to the languages.

• Journal of Intelligence and Information Systems
• /
• v.24 no.4
• /
• pp.51-66
• /
• 2018

As AI (Artificial Intelligence) technologies have been swiftly evolved, a lot of products and services are under development in various fields for better users' experience. On this technology advance, negative effects of AI technologies also have been discussed actively while there exists positive expectation on them at the same time. For instance, many social issues such as trolley dilemma and system security issues are being debated, whereas autonomous vehicles based on artificial intelligence have had attention in terms of stability increase. Therefore, it needs to check and analyse major social issues on artificial intelligence for their development and societal acceptance. In this paper, multi-categorical sentiment analysis is conducted over online public opinion on artificial intelligence after identifying the trending topics related to artificial intelligence for two years from January 2016 to December 2017, which include the event, match between Lee Sedol and AlphaGo. Using the largest web portal in South Korea, online news, news headlines and news comments were crawled. Considering the importance of trending topics, online public opinion was analysed into seven multiple sentimental categories comprised of anger, dislike, fear, happiness, neutrality, sadness, and surprise by topics, not only two simple positive or negative sentiment. As a result, it was found that the top sentiment is "happiness" in most events and yet sentiments on each keyword are different. In addition, when the research period was divided into four periods, the first half of 2016, the second half of the year, the first half of 2017, and the second half of the year, it is confirmed that the sentiment of 'anger' decreases as goes by time. Based on the results of this analysis, it is possible to grasp various topics and trends currently discussed on artificial intelligence, and it can be used to prepare countermeasures. We hope that we can improve to measure public opinion more precisely in the future by integrating empathy level of news comments.