• Architectural research
• /
• 제26권1호
• /
• pp.1-12
• /
• 2024

This study introduces a deep reinforcement learning (DRL)-based methodology for optimizing apartment space arrangements, addressing the limitations of human capability in evaluating all potential spatial configurations. Leveraging computational power, the methodology facilitates the autonomous exploration and evaluation of innovative layout options, considering architectural principles, legal standards, and client re-quirements. Through comprehensive simulation tests across various apartment types, the research demonstrates the DRL approach's effec-tiveness in generating efficient spatial arrangements that align with current design trends and meet predefined performance objectives. The comparative analysis of AI-generated layouts with those designed by professionals validates the methodology's applicability and potential in enhancing architectural design practices by offering novel, optimized spatial configuration solutions.

• 제어로봇시스템학회:학술대회논문집
• /
• 제어로봇시스템학회 2003년도 ICCAS
• /
• pp.2606-2611
• /
• 2003

Comprehensive study on the control system design for a RTP process has been conducted. The purpose of the control system is to maintain maximum temperature uniformity across the silicon wafer achieving precise tracking for various reference trajectories. The study has been carried out in two stages: thermal balance modeling on the basis of a semi-empirical radiation model, and optimal iterative learning controller design on the basis of a linear state space model. First, we found through steady state radiation modeling that the fourth power of wafer temperatures, lamp powers, and the fourth power of chamber wall temperature are related by an emissivity-independent linear equation. Next, for control of the MIMO system, a state space modeland LQG-based two-stage batch control technique was derived and employed to reduce the heavy computational demand in the original two-stage batch control technique. By accommodating the first result, a linear state space model for the controller design was identified between the lamp powers and the fourth power of wafer temperatures as inputs and outputs, respectively. The control system was applied to an experimental RTP equipment. As a consequence, great uniformity improvement could be attained over the entire time horizon compared to the original multi-loop PID control. In addition, controller implementation was standardized and facilitated by completely eliminating the tedious and lengthy control tuning trial.

• 천문학회보
• /
• 제42권2호
• /
• pp.60.2-61
• /
• 2017

A Convolutional Neural Network(CNN) is one of the well-known deep-learning methods in image processing and computer vision area. In this study, we apply CNN to two kinds of flare forecasting models: flare classification and occurrence. For this, we consider several pre-trained models (e.g., AlexNet, GoogLeNet, and ResNet) and customize them by changing several options such as the number of layers, activation function, and optimizer. Our inputs are the same number of SOHO)/MDI images for each flare class (None, C, M and X) at 00:00 UT from Jan 1996 to Dec 2010 (total 1600 images). Outputs are the results of daily flare forecasting for flare class and occurrence. We build, train, and test the models on TensorFlow, which is well-known machine learning software library developed by Google. Our major results from this study are as follows. First, most of the models have accuracies more than 0.7. Second, ResNet developed by Microsoft has the best accuracies : 0.77 for flare classification and 0.83 for flare occurrence. Third, the accuracies of these models vary greatly with changing parameters. We discuss several possibilities to improve the models.

• 로봇학회논문지
• /
• 제17권1호
• /
• pp.76-85
• /
• 2022

One of the most fundamental challenges when designing controllers for dynamic systems is the adjustment of controller parameters. Usually the system model is used to get the initial controller, but eventually the controller parameters must be manually adjusted in the real system to achieve the best performance. To avoid this manual tuning step, data-driven methods such as machine learning were used. Recently, reinforcement learning became one alternative of this problem to be considered as an agent learns policies in large state space with trial-and-error Markov Decision Process (MDP) which is widely used in the field of robotics. However, on initial training step, as an agent tries to explore to the new state space with random action and acts directly on the controller parameters in real systems, MDP can lead the system safety-critical system failures. Therefore, the issue of 'safe exploration' became important. In this paper we meet 'safe exploration' condition with Control Barrier Function (CBF) which converts direct constraints on the state space to the implicit constraint of the control inputs. Given an initial low-performance controller, it automatically optimizes the parameters of the control law while ensuring safety by the CBF so that the agent can learn how to predict and control unknown and often stochastic environments. Simulation results on a quadrotor UAV indicate that the proposed method can safely optimize controller parameters quickly and automatically.

• 국제학술발표논문집
• /
• The 9th International Conference on Construction Engineering and Project Management
• /
• pp.752-759
• /
• 2022

The instantiation of spaces as a discrete entity allows users to utilize BIM models in a wide range of analyses. However, in practice, their utility has been limited as spaces are erroneously entered due to human error and often omitted entirely. Recent studies attempted to automate space allocation using artificial intelligence approaches. However, there has been limited success as most studies focused solely on the use of geometric features to distinguish spaces. In this study, in addition to geometric features, semantic relations between spaces and elements were modeled and used to improve space classification in BIM models. Graph Convolutional Networks (GCN), a deep learning algorithm specifically tailored for learning in graphs, was deployed to classify spaces via a similarity graph that represents the relationships between spaces and their surrounding elements. Results confirmed that accuracy (ACC) was +0.08 higher than the baseline model in which only geometric information was used. Most notably, GCN was able to correctly distinguish spaces with no apparent difference in geometry by discriminating the specific elements that were provided by the similarity graph.

• 제어로봇시스템학회:학술대회논문집
• /
• 제어로봇시스템학회 1999년도 제14차 학술회의논문집
• /
• pp.292-297
• /
• 1999

Evolving in artificial agent is an extremely difficult problem, but on the other hand, a challenging task. At present the studies mainly centered on single agent learning problem. In our case, we use simulated soccer to investigate multi-agent cooperative learning. Consider the fundamental differences in learning mechanism, existing reinforcement learning algorithms can be roughly classified into two types-that based on evaluation functions and that of searching policy space directly. Genetic Programming developed from Genetic Algorithms is one of the most well known approaches belonging to the latter. In this paper, we give detailed algorithm description as well as data construction that are necessary for learning single agent strategies at first. In following step moreover, we will extend developed methods into multiple robot domains. game. We investigate and contrast two different methods-simple team learning and sub-group loaming and conclude the paper with some experimental results.

• 한국환경과학회지
• /
• 제32권8호
• /
• pp.573-583
• /
• 2023

This study was conducted to verify the effectiveness of improving indoor environmental awareness, relieving stress, and improving learning efficiency in school indoor green space, and suggest desirable ways to develop indoor green space in the future. As part of the study, a survey was conducted among 225 individuals across six schools in a metropolitan area with garden and panel-type indoor gardens inside the school building. The survey comprised the current status and use of indoor green spaces, the perception of indoor green spaces, improvement measures in indoor green spaces, and basic properties. Semantic Differential (SD) was used to investigate the impression of school indoor spaces. Resultantly, the more frequent the use of green spaces in the school, the more they feel the positive effects of indoor green spaces, such as improving the school's indoor environment, reducing stress, and improving learning efficiency. In addition, it appears that the more frequent contact with the natural environment, the more they feel the positive effects provided by indoor green space at school. Therefore, it is suggested that educational conditions must be improved by revitalizing various green welfare, including indoor green areas, at the school level.

• 한국콘텐츠학회논문지
• /
• 제5권2호
• /
• pp.53-60
• /
• 2005

e-teaming은 시간적, 공간적 제약으로부터의 자유스러움으로 인해 전통적 교육의 보조 도구로써, 또는 전통적 교육을 대신해서 이미 많은 부분에서 e-learning이 이루어지고 있다. 본 논문은 e-learning의 중요성과 그것이 차지하는 비중이 상당히 커져 있는 시점에서 e-learning 교육 경험자들이 인터넷 기반 e-learning의 학습효과를 전통적 교육에 비해 어느 정도로 인식하고 있는지에 대한 조사를 수행하였다. 또한 인터넷 기반 e-learning의 학습효과에 대한 요인변수를 도출하고, 인터넷 기반e-loaming의 학습효과로 인식하는 정도가 어떠한 요인에 의해 얼마만큼의 영향을 받고 있는지를 파악하여 제시하였다.

• 의료ㆍ복지 건축 : 한국의료복지건축학회 논문집
• /
• 제15권3호
• /
• pp.71-77
• /
• 2009

This study is to understand the situations of special education classroom layout, find differences according to school levels and summarize the findings in order to build up the indicators for special classroom layout. As for elementary school level, special classrooms are using multi-purposes desk or group desk for diverse activities such as basic learning and formation of basic life practice. The most frequent type in classroom layout is Type E which is for diverse coner-learning and play activities and the next is Type C which secures activity space. Because security of dynamic activity which most teachers find problematic is important, it needs more research to secure dynamic activity space within classroom. As for middle school level, the most frequent type in classroom layout is Type B which is equiped for computer aided learning and the next is Type C which secures activity space. Research for systematic layout of activity space is needed in order to secure the spaces of dynamic activity and basic job training. As for high school levels, mostly Type B which emphasizes computer activities is adopted and next is Type F which is capable for job training. The survey about the size of special education classroom proves that most teachers want one and half size classroom which in not such a large classroom. It is expected that more systematic research of special classroom layout according to school levels may reach for rational space layout.

• 교육시설 논문지
• /
• 제23권1호
• /
• pp.43-54
• /
• 2016

The purpose of this study was to survey teachers' perception of classroom space utilization through analyzing the physical environment of elementary school classrooms (56 classrooms in 10 schools). Most of the teachers arranged desks in the two person parallel type (sectional layout : standard type) for their classes. Although the number was small, some classrooms used the T type, H type, U type, group type, and the teachers of such cases used these layouts for children's play activities or group learning. Some teachers changed the desk layout depending on the contents of learning or for different atmosphere of class, but about 40% of the teachers used the same classroom layout without any change during a semester. When the teachers' perception of classroom space utilization was examined according to the type and change of desk layout, the quantity and characteristics of posts, the position of posting spaces, and the size of activity spaces in the classroom, most of the teachers tended to be conventional without any characteristic, and only 16% of them were relatively active in utilizing classroom spaces. In addition, teachers of a relatively small class were more active in utilizing classroom spaces. In particular, perception was very low to utilize the classroom as a space for children's life or play activities or various types of learning. These findings suggest that it is necessary to improve teachers' perception of classroom space utilization in the future.