간행물

중앙대학교 인문콘텐츠연구소> 인공지능인문학연구

인공지능인문학연구 update

Journal of AI Humanities (JAIH)

  • : 중앙대학교 인문콘텐츠연구소
  • : 인문과학분야  >  기타(인문과학)
  • :
  • :
  • : 연속간행물
  • : 반년간
  • : 2635-4691
  • :
  • :

수록정보
수록범위 : 1권0호(2018)~4권0호(2019) |수록논문 수 : 28
인공지능인문학연구
4권0호(2019년 10월) 수록논문
최근 권호 논문
| | | |

1Homo Faber -The Impact of Tool Use Cultures on Ecosystems and Society

저자 : Hans-martin Sass

발행기관 : 중앙대학교 인문콘텐츠연구소 간행물 : 인공지능인문학연구 4권 0호 발행 연도 : 2019 페이지 : pp. 9-30 (22 pages)

다운로드

(기관인증 필요)

초록보기

We humans are creating and using tools-hardware tools and software tools, and we love to play:we correctly have called ourselves Homo Faber and Homo Ludens, the species which uses tools and loves to play. Hardware tools include hammers and axes, farms and vineyards, houses cars and mails, networks of electricity and commerce. Software tools include rules and values, punishments and honors, Gods and Goddesses, good and evil spirits, languages and traditions, ceremonies and festivals, parties and plays. I present a Biocultural Imperative:'Support direct human inter-action and common-sense as an end in itself and use robots and other artificial and empathy intelligence as tools only in stabilizing human cultures, enjoyable and open to the future'.

2Philosophy and the Politics of Moral Machines

저자 : Paul Dumouchel

발행기관 : 중앙대학교 인문콘텐츠연구소 간행물 : 인공지능인문학연구 4권 0호 발행 연도 : 2019 페이지 : pp. 31-50 (20 pages)

다운로드

(기관인증 필요)

초록보기

A recent large-scale survey, “The Moral Machine experiment” (2018) aggregated 39.61 million decisions across 233 countries and territories reflecting people's preferences as to who should be spared in fatal moral dilemmas involving autonomous road vehicles. The experiment collected 'big data' to reach conclusions concerning the moral rules that should be implemented in these vehicles. In this paper, first I question the philosophical presuppositions of the experiment, arguing that it has very little to do with ethics or moral norms, but essentially constitutes a market survey concerning the social acceptance of a dangerous technology. Then, I criticize the myth of moral machines and the illusion that abandoning to automated systems the power to 'autonomously' take lethal 'decisions' is a radically new phenomenon. Finally, I suggest a different solution to the difficulties addressed by the Moral Machine experiment and make political and legal suggestions concerning policy towards 'autonomous road vehicles'.

3Use of Artificial Intelligence in Sustainable Agriculture -A Preliminary Analysis

저자 : Egemen Bezci

발행기관 : 중앙대학교 인문콘텐츠연구소 간행물 : 인공지능인문학연구 4권 0호 발행 연도 : 2019 페이지 : pp. 51-79 (29 pages)

다운로드

(기관인증 필요)

초록보기

The use of artificial intelligence (AI) has great potentials to address the pressing sustainability issues in the Anthropocene epoch. One of the most important use of AI technologies is recognized as its role to increase efficiency of the agricultural production to address poverty, and to reduce the negative environmental impacts. One of the important steps, thus, is to reduce the agricultural greenhouse gas emissions. However, an expert survey conducted as a part of this study warned that the AI applications for agricultural sector could creatr perverse incentives, increase hidden energy costs and ignore the externalities. This article initially surveys the role of the Artificial Intelligence for the agricultural production. Following the initial literature survey, this research provides a basic preliminary empirical data analysis from 73 countries to demonstrate the whether their advancements in AI applications has decoupled their Agricultural Greenhouse gas emissions from the food production.

4Deep Learning -Shifting the Telos from Singularity to Perception

저자 : Aditya Nayak

발행기관 : 중앙대학교 인문콘텐츠연구소 간행물 : 인공지능인문학연구 4권 0호 발행 연도 : 2019 페이지 : pp. 81-99 (19 pages)

다운로드

(기관인증 필요)

초록보기

This paper seeks to undertake a philosophical enquiry of Artificial Intelligence. The paper specifically deals with Deep Reinforcement Learning, and Neural Networks. The first half of the paper shall discuss the ontology of the field of Artificial Intelligence where the discussion is primarily on the methodology used for AI while trying to locate it in the space of scientific evolution in the vocabulary of smooth and striated Space, as conceptualized by Gilles Deleuze and Felix Guattari. The second part of the presentation undertakes a philosophical understanding of the process of AI, i.e. the three sequential steps in developing an AI program for iterative learning and implementation, namely- representation, evaluation and optimization. This shall prepare the discussion for the vocabulary of 'virtuality', and 'perception'; concepts given by Henri Bergson, and Deleuze and developed further by Brian Massumi. The presentation concludes with a prescriptive argument towards solving the problem faced by Deep Learning in understanding human subjectivity, reason, emotions, and how this can be resolved by changing the telos from 'singularity' to 'perception'.

5Can Machines Play Language Games with Humans?

저자 : Bingnan Zheng , Tianqun Pan

발행기관 : 중앙대학교 인문콘텐츠연구소 간행물 : 인공지능인문학연구 4권 0호 발행 연도 : 2019 페이지 : pp. 101-114 (14 pages)

다운로드

(기관인증 필요)

초록보기

Since the introduction of the Turing test and the emergence of the concept of artificial intelligence, the performance of machines in “dialogue” with humans has been an important criterion for judging machine intelligence. Machine intelligence is embodied in the use of language. Wittgenstein advocated, in his theory of language games, that the meaning of language lies in its use, and that language cannot be completely separated from human life activities. Language has several fixed rules, but there are no specific rules in many aspects. Furthermore, language games and life forms are interdependent. Therefore, under a lower standard, machines can play limited and rule-based language games with humans, thus achieving specialized intelligence, but they cannot fully participate in human life practices; therefore, they have not achieved universal intelligence. However, to seek a broader development space for artificial intelligence, one must leave the theoretical framework of anthropocentrism. By following this approach, a future world of human- machine coexistence can be expected.

6Open Knowledge Graph of Historical Heritages for Korean Religions

저자 : Moon Hee-kyung , Zhanfang Zhao , Kim Jae-ik , Park Kwang-soo

발행기관 : 중앙대학교 인문콘텐츠연구소 간행물 : 인공지능인문학연구 4권 0호 발행 연도 : 2019 페이지 : pp. 115-134 (20 pages)

다운로드

(기관인증 필요)

초록보기

The Knowledge Graphs (KGs) as the emerging artificial intelligence (AI) paradigm play a crucial role in knowledge representing and sharing to achieve intelligent knowledge services. This paper addresses a new Linked Open Data (LOD) development method that can realize KGs of the diverse domains from the practical perspectives. The proposed approach is based on two types of separation of concerns for the effective development of LOD data sets:conceptual layer separation between schema and instances, and view separation between LOD management system and user interface. In order to mediate separated concerns, a LOD Interface Sheet (LIS) is implemented to create a domain-specific LOD development environment. The paper implements the proposed method that can generate the domain-specific LOD development system providing the necessary functionalities for LOD management as well as KG realization. This paper also presents the development of an open knowledge graph of historical heritages for Korean religions to demonstrate the capability of the implemented i-Manager and the effectiveness of KGs in scholarly knowledge management.

7딥러닝 기반 자연어 처리에서 도메인 지식의 역할

저자 : 박진호 ( Park Jin-ho )

발행기관 : 중앙대학교 인문콘텐츠연구소 간행물 : 인공지능인문학연구 4권 0호 발행 연도 : 2019 페이지 : pp. 135-166 (32 pages)

다운로드

(기관인증 필요)

초록보기

Symbolic AI에서는 도메인 지식이 중요시되었다. 규칙 기반 자연어처리에서도 언어학적 지식이 중요한 역할을 담당했다. 확률 기반 자연어처리와 기계학습기법이 발달하면서 도메인 지식의 역할은 축소되었다. 딥러닝이 대두하면서, 자질 공학과 도메인 지식의 역할은 훨씬 더 축소되었다. 딥러닝 시대에도 여전히 도메인 전문가(언어학자)의 역할이 중요함을 증명하기 위해 한국어 형태소분석기를 개발하였다. 한국어는 형태음소적 교체, 탈락, 축약이 활발하여 분절 과제가 쉽지 않지만, 분절 과제를 분류 문제로 재설정하면 기계학습으로 더 쉽게 해결할 수 있게 된다. 이를 위해서는 분절 이전의 입력의 각 음절과 분절된 출력의 대응하는 문자열 사이의 매핑 관계를 망라적으로 목록화하는 것이 관건이다. 1200만 어절 규모의 세종 형태의미 분석 말뭉치를 통해 이러한 매핑에 200개 유형이 있음을 확인하였다. 이 200개 범주를 바탕으로 LSTM 기반 신경망 모델을 만들어 훈련시켰다. 분절 문제가 해결되면, 분절된 각 토큰에 대한 레이블링은, 영어 등에 대한 선행 연구로 친숙한 연쇄 레이블링 알고리즘으로 쉽게 해결할 수 있다. 이 두 가지 모델과 사전을 결합하여, F1 스코어 98.0%의 성능을 얻을 수 있었다. 이 실험은 딥러닝 시대에도 도메인 지식이 여전히 중요함을 보여준다.


In Symbolic AI, the domain knowledge was considered indispensable. In rule-based NLP, likewise, the linguistic knowledge played an important role. As probabilistic NLP and machine learning techniques develop, the role of domain knowledge shrank. As deep learning appears, even the role of feature engineering and domain knowledge has become almost zero.
In order to prove the importance of domain knowledge even in this deep learning age, I built a parts-of-speech tagger of Korean. This task in Korean is challenging, due to morphophonological alternations, deletions and contractions. I reformulated this task of segmentation as that of classification. For this purpose, I examined a large corpus, and found empirically 200 types of mapping between an input syllable and an output string. Based on these categories, I built and trained an LSTM-based neural network. With this model of segmentation, the parts-of-speech tagging model is easily trained by the familiar sequence tagging algorithm. By combining these two models and a few dictionaries, I got 98.0% of the F1 score.

1
권호별 보기
가장 많이 인용된 논문

(자료제공: 네이버학술정보)

가장 많이 인용된 논문
| | | |
1연안해역에서 석유오염물질의 세균학적 분해에 관한 연구

(2006)홍길동 외 1명심리학41회 피인용

다운로드

2미국의 비트코인 규제

(2006)홍길동심리학41회 피인용

다운로드

가장 많이 참고한 논문

(자료제공: 네이버학술정보)

가장 많이 참고한 논문

다운로드

2미국의 비트코인 규제

(2006)홍길동41회 피인용

다운로드

해당 간행물 관심 구독기관

전북대학교 호서대학교 울산대학교 이화여자대학교 숙명여자대학교
 46
 27
 21
 15
 13
  • 1 전북대학교 (46건)
  • 2 호서대학교 (27건)
  • 3 울산대학교 (21건)
  • 4 이화여자대학교 (15건)
  • 5 숙명여자대학교 (13건)
  • 6 연세대학교 (9건)
  • 7 대구대학교 (9건)
  • 8 중앙대학교(서울) (8건)
  • 9 한국외국어대학교 (7건)
  • 10 동국대학교 (7건)

내가 찾은 최근 검색어

최근 열람 자료

맞춤 논문

보관함

내 보관함
공유한 보관함

1:1문의

닫기