• Title/Summary/Keyword: 관계형 강화 학습

Search Result 20, Processing Time 0.024 seconds

Using Prior Domain Knowledge for Efficient Relational Reinforcement Learning (효율적인 관계형 강화학습을 위한 사전 영역 지식의 활용)

  • Kang, Minkyo;Kim, Incheol
    • Proceedings of the Korea Information Processing Society Conference
    • /
    • 2021.05a
    • /
    • pp.483-486
    • /
    • 2021
  • 기존의 심층 강화학습은 상태, 행동, 정책 등을 모두 벡터 형태로 표현하는 강화학습으로서, 학습된 정책의 일반성과 해석 가능성에 제한이 있고 영역 지식을 학습에 효과적으로 활용하기도 어렵다는 한계성이 있다. 이러한 문제점들을 해결하기 위해 제안된 새로운 관계형 강화학습 프레임워크인 dNL-RRL은 상태, 행동, 그리고 학습된 정책을 모두 논리 서술자와 규칙들로 표현할 수 있다. 본 논문에서는 dNL-RRL을 기초로 공장 내 운송용 모바일 로봇의 제어를 위한 행동 정책 학습을 수행하였으며, 학습의 효율성 향상을 위해 인간 전문가의 사전 영역 지식을 활용하는 방안들을 제안한다. 다양한 실험들을 통해, 본 논문에서 제안하는 영역 지식을 활용한 관계형 강화학습 방법의 학습 성능 개선 효과를 입증한다.

Effective Utilization of Domain Knowledge for Relational Reinforcement Learning (관계형 강화 학습을 위한 도메인 지식의 효과적인 활용)

  • Kang, MinKyo;Kim, InCheol
    • KIPS Transactions on Software and Data Engineering
    • /
    • v.11 no.3
    • /
    • pp.141-148
    • /
    • 2022
  • Recently, reinforcement learning combined with deep neural network technology has achieved remarkable success in various fields such as board games such as Go and chess, computer games such as Atari and StartCraft, and robot object manipulation tasks. However, such deep reinforcement learning describes states, actions, and policies in vector representation. Therefore, the existing deep reinforcement learning has some limitations in generality and interpretability of the learned policy, and it is difficult to effectively incorporate domain knowledge into policy learning. On the other hand, dNL-RRL, a new relational reinforcement learning framework proposed to solve these problems, uses a kind of vector representation for sensor input data and lower-level motion control as in the existing deep reinforcement learning. However, for states, actions, and learned policies, It uses a relational representation with logic predicates and rules. In this paper, we present dNL-RRL-based policy learning for transportation mobile robots in a manufacturing environment. In particular, this study proposes a effective method to utilize the prior domain knowledge of human experts to improve the efficiency of relational reinforcement learning. Through various experiments, we demonstrate the performance improvement of the relational reinforcement learning by using domain knowledge as proposed in this paper.

Analysis of Educational Effects in Augmented Reality Combined Marker System (증강현실 조합형 마커시스템의 교육효과분석)

  • Ko, Youngnam;Kim, Chongwoo
    • Journal of The Korean Association of Information Education
    • /
    • v.16 no.3
    • /
    • pp.373-382
    • /
    • 2012
  • Of computing skills in the field of multi-media, particularly augmented reality technology contents may provide realistic learning experiences with 3D pictures through the learners' manipulation activities. However, the marker systems in the existing studies were not well developed as to maintain the students' interest and concentration. In this study, we have designed the first lesson ("Earth and Moon") of 5th graders' science with augmented reality combined system so that we could deal with manipulation activities of the relationship between augmented objects, From the experimental study, using combined augmented reality contents made a significant difference in their learning achievement and motivation. Thus augmented reality combined system can be utilized for a variety of topics to maintain students' learning motivation.

  • PDF

대학의 학생창업에 미치는 창업동아리의 역할

  • Won, Chi-Un;Bae, Tae-Jun
    • 한국벤처창업학회:학술대회논문집
    • /
    • 2019.11a
    • /
    • pp.87-93
    • /
    • 2019
  • 본 연구는 최근 창업에 대한 관심이 대학 내에서 확산되고 성공한 학생창업가들이 배출되고 있는 시점에서 대학 내 창업동아리 활동이 실제로 학생창업으로 이어지는지 국내 4년제 대학 169개를 대상으로 실증분석을 실시하였다. 대학의 창업동아리는 학생창업을 활성화하기 위한 방안으로써 정부와 대학은 매년 창업동아리에 필요한 예산과 교육 등 적극적인 지원을 하고 있지만 창업동아리 활동이 학생창업에 어떻게 영향을 미치는지 국내에서는 학술적으로 연구된 바가 극히 드물다. 이러한 이유로 본 연구에서는 대학의 창업동아리가 학생창업 증가에 미치는 영향을 분석하고자 한다. 구체적으로 선행연구에서 강조한 바와 같이 창업동아리 활동이 학생창업에 미치는 영향을 대학의 창업지원과 교육의 조절효과를 중심으로 실증분석 한다. 분석결과 첫째, 대학의 창업동아리가 실제로 학생창업에 정(+)의 영향을 미치는 것으로 나타났다. 이러한 결과는 창업동아리 활동이 활발한 대학일수록 학생창업이 많이 이루어진다는 것을 시사한다. 둘째, 학생 창업지원과 교육의 조절효과를 검증한 결과 대학의 실습형 창업강좌는 창업동아리 활동과 학생창업 간의 관계를 정(+)의 방향으로 조절한다는 증거가 발견되었다. 즉, 창업동아리 활동이 학생창업에 미치는 긍정적인 영향은 실습형 창업강좌의 비중이 높은 학교일수록 긍정적인 관계가 강화된다는 것을 의미한다. 이는 경험적 학습을 통한 동아리 활동과 실습형 교육 간 연계의 필요성을 강조한다. 따라서, 창업 동아리 활동을 통한 학생들의 창업활동을 분석함으로써 대학의 창업 동아리 활동의 효과를 증진시킬 수 있는 현실적인 대안을 제시하고자 한다. 본 연구의 결과는 대학 내 활발한 창업 동아리 활동의 질적 향상과 실제 창업 동아리 활동을 경험한 학생들이 창업으로 이루어질 수 있는 현실적인 창업 교육에 기여할 수 있는 것으로 기대된다.

  • PDF

An Analysis of Structural Relationship between Technological Innovation Capability, Collaboration and New Product Development Performance in Small & Mid-sized Venture Companies (중소벤처기업의 기술혁신역량, 협업, 신제품개발성과 간의 구조적 관계 분석)

  • Lee, Rok
    • Asia-Pacific Journal of Business Venturing and Entrepreneurship
    • /
    • v.15 no.1
    • /
    • pp.185-195
    • /
    • 2020
  • This study is intended to determine that there is a casual relationship between technological innovation capability and new product development performance in small and mid-sized venture companies, and that the introduction of collaboration as a means to step up technological innovation capability will improve new product development performance. To achieve this, a survey was carried out to employees who are engaged in R&D work for small and mid-sized venture companies based in Korea, and the results were analyzed by regression analysis. The findings showed that technology strategy, technology learning and open innovation belonging to technological innovation capability in small and mid-sized venture companies had an effect on new product development performance. In other words, the selection of collaboration as a wider array of core strategies on new product development performance showed that collaboration was a strategy affecting new product development performance. In addition, the moderating role of technological innovation capability in boosting new product development performance through the introduction of collaboration showed that common collaboration had a positive effect on stepping up technology strategy, and collaboration as a core strategy had a positive effect on the size of new product development performance by strengthening technology strategy and open innovation.

The multi agent control heuristic using direction vector (방향 벡터를 이용한 다중에이전트 휴리스틱)

  • Kim Hyun;Lee SeungGwan;Chung TaeChoong
    • Proceedings of the Korea Information Processing Society Conference
    • /
    • 2004.11a
    • /
    • pp.525-528
    • /
    • 2004
  • 먹이추적문제(prey pursuit problem)는 가상 격자로 이루어진 공간 내에 다중의 에이전트를 이용하여 먹이를 포획하는 것이다. 에이전트들은 먹이를 포획하기 위해 $30{\times}30$으로 이루어진 격자공간 (gride)안에서 기존 제안된 지역 제어, 분산 제어, 강화학습을 이용한 분산 제어 전략들을 적용하여 먹이를 포획하는 전략을 구현하였다. 제한된 격자 공간은 현실세계를 표현하기에는 너무도 역부족이어서 본 논문에서는 제한된 격자공간이 아닌 현실 세계와 흡사한 무한 공간 환경을 표현하고자 하였다. 표현된 환경의 모델은 순환구조(circular)형 격자 공간이라는 새로운 실험 공간이며, 새로운 공간에 맞는 전략은 에이전트와 먹이와의 추적 관계를 방향 벡터를 고려한 모델로 구현하였다. 기존 실험과는 차별화 된 환경에서 에이전트들은 휴리스틱을 통한 학습을 할 수 있다는 가정과 먹이의 효율적 포획, 충돌문제 해결이라는 결과를 얻었다.

  • PDF

Capacity Building Programs for Emerging Countries by the Korean Regional Innovation Model: Policy Analysis and Suggestions (한국형 지역혁신모델의 신흥국 전수사업 : 정책분석과 제안)

  • Kim, Hak-Min
    • Journal of the Korea Academia-Industrial cooperation Society
    • /
    • v.19 no.3
    • /
    • pp.75-82
    • /
    • 2018
  • Recently, emerging countries have been paying attention to Korean economic development policy, trying to adopt the Korean regional innovation model. Korea is also interested in exporting its regional innovation model and enhancing economic cooperation with those countries. This paper aims to analyze the capacity-building programs of the Korean regional innovation model for emerging countries and suggests policies for it. For this purpose, the local innovators' participation patterns in the process of collaborative learning/networking/interaction are investigated with a focused group-interview method. From an analysis of the programs supported by Korean organizations, this study finds that the correlation coefficient between the training time of capacity building and the participation rate of local members' collaborative learning is very high (0.975). Since the correlation coefficient between the participation rates of collaborative learning and networking is relatively low (0.667), a policy to link local collaborative learning to networking should be provided. As the correlation coefficient between the participation rates of networking and interaction is high (0.950), networking is a key to regional innovation. This study recommends activity programs to promote networking among local innovators, rather than training and consulting programs. As introduced in the Chungnam Techno Park case, this study suggests that the capacity-building program should include programs to initiate a collaborative learning network, to create a local-demand, regional innovation model, and to operate the regional innovation platform, which should be done by local innovators in the emerging countries.

Effectiveness and Relationship Analysis of Chemistry Programs Based on Metacognitive Learning Strategies Using Realistic Contents for Pre-service Teachers (예비교사를 위한 실감형 콘텐츠 활용 메타인지 학습전략 기반 화학 프로그램의 효과 및 관계성 분석)

  • Da Eun Lee;Hyun-Kyung Kim
    • Journal of the Korean Chemical Society
    • /
    • v.67 no.4
    • /
    • pp.271-280
    • /
    • 2023
  • The purpose of this study is to investigate the effect of chemistry program based on metacognitive learning strategies using realistic contents on prospective teachers' creative thinking skills and science core competencies, and their perception. In particular, it was intended to further improve the effectiveness of the program by introducing a strategy to strengthen metacognition. Participants were classified into the experimental group subject to the newly developed chemistry curriculum and traditional group subject to general programs that exclude realistic contents and metacognitive strategies. Both groups were surveyed before and after the application of the program to measure the degree of change in metacognitive competencies, creative thinking competencies, and science core competencies. It also analyzed the impact of metacognitive competencies and science core competencies on creativity thinking competencies. As a result of the study, relevance and rationality among sub-factors of metacognitive competencies and creative thinking competencies of the experimental group were improved, and all sub-factors except for scientific participation and lifelong learning ability among science core competencies were significantly improved. In addition, it was found that metacognitive knowledge among metacognitive competencies, scientific inquiry ability and scientific thinking ability among science core competencies affect creative thinking competencies. Through the results, it was suggested that realistic content that incorporates metacognitive learning strategies is needed to improve creative thinking competencies, and learning models and programs that can utilize them are needed.

Influencing Factors the Metacognition and Learning Motivation on Problem-Solving Ability in Nursing Students (간호대학생의 메타인지, 학습 동기가 문제해결력에 미치는 영향요인)

  • Kim, Nam Young
    • Journal of the Korean Applied Science and Technology
    • /
    • v.38 no.4
    • /
    • pp.931-940
    • /
    • 2021
  • The purpose of this research is to identify how the metacognition and learning motivation of nursing students affect problem-solving ability. The subjects of this study were 160 students attending nursing universities, and the data were collected from 20 January to 10 March 2021. Data were analyzed using the SPSS/WIN 24.0 program. As a result of this study, there was a significant positive correlation between problem-solving ability, metacognition (r=.44, p<.001), and learning motivation(r=.45, p<.001). factors affecting the problem-solving ability of nursing students were identified in order of learning motivation, metacognition. Based on the results of this study, it is hoped that programs will be developed and applied to improve the problem-solving ability of nursing students.

Real-Time Scheduling Scheme based on Reinforcement Learning Considering Minimizing Setup Cost (작업 준비비용 최소화를 고려한 강화학습 기반의 실시간 일정계획 수립기법)

  • Yoo, Woosik;Kim, Sungjae;Kim, Kwanho
    • The Journal of Society for e-Business Studies
    • /
    • v.25 no.2
    • /
    • pp.15-27
    • /
    • 2020
  • This study starts with the idea that the process of creating a Gantt Chart for schedule planning is similar to Tetris game with only a straight line. In Tetris games, the X axis is M machines and the Y axis is time. It is assumed that all types of orders can be worked without separation in all machines, but if the types of orders are different, setup cost will be incurred without delay. In this study, the game described above was named Gantris and the game environment was implemented. The AI-scheduling table through in-depth reinforcement learning compares the real-time scheduling table with the human-made game schedule. In the comparative study, the learning environment was studied in single order list learning environment and random order list learning environment. The two systems to be compared in this study are four machines (Machine)-two types of system (4M2T) and ten machines-six types of system (10M6T). As a performance indicator of the generated schedule, a weighted sum of setup cost, makespan and idle time in processing 100 orders were scheduled. As a result of the comparative study, in 4M2T system, regardless of the learning environment, the learned system generated schedule plan with better performance index than the experimenter. In the case of 10M6T system, the AI system generated a schedule of better performance indicators than the experimenter in a single learning environment, but showed a bad performance index than the experimenter in random learning environment. However, in comparing the number of job changes, the learning system showed better results than those of the 4M2T and 10M6T, showing excellent scheduling performance.