• Bulletin of the Korean Chemical Society
• /
• 제33권2호
• /
• pp.613-619
• /
• 2012

We applied several machine learning methods for developing QSAR models for prediction of acute toxicity to fathead minnow. The multiple linear regression (MLR) and artificial neural network (ANN) method were applied to predict 96 h $LC_{50}$ (median lethal concentration) of 555 chemical compounds. Molecular descriptors based on 2D chemical structure were calculated by PreADMET program. The recursive partitioning (RP) model was used for grouping of mode of actions as reactive or narcosis, followed by MLR method of chemicals within the same mode of action. The MLR, ANN, and two RP-MLR models possessed correlation coefficients ($R^2$) as 0.553, 0.618, 0.632, and 0.605 on test set, respectively. The consensus model of ANN and two RP-MLR models was used as the best model on training set and showed good predictivity ($R^2$=0.663) on the test set.

• 한국HCI학회:학술대회논문집
• /
• 한국HCI학회 2006년도 학술대회 2부
• /
• pp.518-523
• /
• 2006

As digital broadcasting process and information compressing technologies developed, digital technologies have been used extensively in many visual media such as movies and TV programs, and it is required to develop a new process for 3 dimensional digital programs. The study attempts to develop a new digital video-producing process through pre-simulation using a digital storytelling method, in order to increase the quality of broadcasting, reducing the producing time and improve efficiency. This system, supposed to be used to produce actual broadcasting, includes all tools needed for broadcasting in the program, so users can select an appropriate set among the pre-made library, which saves time for making sets. Also, it offers a storyboard making function by perfectly representing the functions of a virtual camera, and enables users to make the better storyboard, checking the real time movies. In addition, it recognizes objects though avatars, organizes the appropriate set for these avatars to act, and embodies a variety of avatar movements through which it reaffirms the camera's position and organization, minimizing errors and saving time and budget.

• 화약ㆍ발파
• /
• 제28권2호
• /
• pp.1-16
• /
• 2010

본 연구에서는 암반사면형성 작업시 최종벽면에 적용된 프리스플리팅(pre-splitting) 발파공법의 발파효과와 관련하여 발파 후 벽면의 암반손상이 장약된 폭약의 폭력 보다는 암반 내에 발달된 불연속면의 발달 형태에 따라 더 큰 영향을 보일 수 있음을 규명하였다. 이를 위하여 불연속면을 대표할 수 있는 절리에 대한 조사를 통해 발파 후 벽면 암반의 절리군 분포양상을 4가지 Case로 분류하고, 파쇄도 분석 영상처리시스템을 통해 벽면에 나타나는 암반블록의 크기 빈도를 비교 분석함으로써 벽면의 암반손상도를 파악하였다. 절리군이 1개 이하로 발달하는 경우, 분석된 블록의 크기 중 2,000mm 이상 되는 부분이 42%를 차지하여 프리스플리팅 발파공법의 효과를 뚜렷이 확인할 수 있었으며, 2~3개의 절리군이 일방향으로 발달하는 경우와 교차되면서 발달하는 경우, 블록의 크기는 1,000~2,000mm 사이에 각각 43.6% 및 35.8%의 빈도로 분포하는 것으로 나타나 프리스플리팅 발파공법에 의한 발파효율이 다소 떨어지는 양상을 보였다. 그러나 3개 이상의 절리군이 불규칙하게 발달하는 경우에는 블록의 크기가 250~500mm 사이에 35%의 빈도로 분포하고 1,000mm 이상의 크기에 대해서는 거의 나타나지 않는 양상을 보였다. 따라서 이러한 경우 프리스플리팅 발파공법에 의한 발파 효과는 거의 없이 일반적인 발파가 이루어졌다고 볼 수 있었다. 또한 PFC2D에 의한 발파수치해석결과, 암반 내부로의 손상영역 발생은 본 발파보다는 프리스플리팅 발파공법에 의해 직접적인 영향을 받을 수 있음을 확인하였으며, 따라서 향후 사면 형성을 위한 프리스플리팅 발파공법을 적용할 경우에는 사전 지표지질조사를 시행하여 절리를 비롯한 불연속면과 관련된 사항을 충분히 파악할 필요가 있으며, 시공진행에 따라 예상보다 많은 절리군이 나타날 경우에는 프리스플리팅 발파공법의 설계 조정이 필요할 것으로 판단된다.

• 한국전산구조공학회논문집
• /
• 제27권4호
• /
• pp.313-320
• /
• 2014

텐세그리티 구조물은 인장력을 받는 연속된 케이블 안에 압축력을 받는 스트럿이 결합된 형태로 구성된다. 텐세그리티 구조물은 자기 응력 상태를 갖는 프리스트레스 핀 접합 구조물에 속한다. 텐세그리티 구조물 설계의 핵심은 평형 배열상태를 구하는 일명 형상탐색 과정이다. 본 논문에서는 세 가지의 효과적인 텐세그리티 구조물의 형상탐색 기법을 제안하였다. 형상탐색과정을 수행하면 평형상태의 내력 밀도와 그에 대응하는 위상을 얻을 수 있다. 이 때 평형상태를 형성하는 적절한 내력밀도 값을 얻기 위해 유전자 알고리즘을 결합한 내력밀도법이 사용되었다. 수치해석 예제를 통해 제안 알고리즘의 효율성을 입증하였다.

• 지능정보연구
• /
• 제9권1호
• /
• pp.251-267
• /
• 2003

객체지향 데이터 베이스에서는 지능 정보시스템에서 요구하는 것을 만족하기 위하여 보다 지능적인 질의 처리 기법이 필요하다. 본 논문에서는 추론적 기법을 사용하여, 객체지향 데이터베이스에서의 지능적인 질의 처리하는 방법에 대하여 논의한다. 논문에서 제시하는 방법을 사용하여, 객체지향 데이터베이스에서 주어진 질의에 대한 답을 추상적으로 표현하는 지능적인 답을 얻을 수 있다. 본 논문에서 제안하는 지능적인 질의 처리 방법은 규칙 표현, 규칙 재편성, 전 분석, 분석의 네 단계로 구성된다. 규칙 표현 단계에서는 객체지향 데이터베이스 스키마를 사용하여 추론 규칙을 생성한다. 규칙 재편성 단계에서는 규칙에서 순환을 제거한다. 전 분석 단계에서는 유일한 내포적 문자를 얻기 위하여 규칙변환이 이루어진다. 분석 단계에서는 SLD-분석을 사용하여 내포적 답을 구한다.

• ETRI Journal
• /
• 제40권4호
• /
• pp.435-445
• /
• 2018

This paper proposes a human-like sign-language learning method that uses a deep-learning technique. Inspired by the fact that humans can learn sign language from just a set of pictures in a book, in the proposed method, the input data are pre-processed into an image. In addition, the network is partially pre-trained to imitate the preliminarily obtained knowledge of humans. The learning process is implemented with a well-known network, that is, a convolutional neural network. Twelve sign actions are learned in 10 situations, and can be recognized with an accuracy of 99% in scenarios with low-cost equipment and limited data. The results show that the system is highly practical, as well as accurate and robust.

• Steel and Composite Structures
• /
• 제48권1호
• /
• pp.73-88
• /
• 2023

This paper presents a novel implicit level set method for topology optimization of functionally graded (FG) structures with pre-existing discontinuities (pre-cracks) using radial basis functions (RBF). The mathematical formulation of the optimization problem is developed by incorporating RBF-based nodal densities as design variables and minimizing compliance as the objective function. To accurately capture crack-tip behavior, crack-tip enrichment functions are introduced, and an eXtended Finite Element Method (X-FEM) is employed for analyzing the mechanical response of FG structures with strong discontinuities. The enforcement of boundary conditions is achieved using the Hamilton-Jacobi method. The study provides detailed mathematical expressions for topology optimization of systems with defects using FG materials. Numerical examples are presented to demonstrate the efficiency and reliability of the proposed methodology.

• 한국축산식품학회지
• /
• 제42권6호
• /
• pp.942-952
• /
• 2022

To establish a pre-plating method of chicken satellite cells with high purity, pre-plating was performed under culture conditions of 37℃ and 41℃, and the pre-plating time was set from a total of 3 hours to 6 hours in consideration of the cell attachment time. The purity of the cells was confirmed by staining paired box protein 7 (Pax7) after proliferation, and Pax7 expression was the highest in culture flasks shaken for 2 hours after incubation at 41℃ for 2 hours to prevent the attachment of satellite cells (p<0.05). Also, when pre-plating and proliferation were performed at 37℃ and 41℃, the Pax7 expression rate was higher at 41℃. The differentiation capabilities of the three groups (T3, T6, and T7) with high Pax7 expression were compared and the fusion index (%) and myotube formation area (%) determined by myosin heavy chain (MHC) staining was calculated. The T6 and T7 groups, which were cultured at 41℃, showed significantly higher values than the T3 group (p<0.05). There was no significant difference in the expression of Pax7 and MHC between the T6 and T7 groups (p>0.05). These results suggest that pre-plating at 41℃ for a total of 4 hours was the most efficient in terms of cost and time for purifying chicken satellite cells for cultured meat.

• 한국초등과학교육학회지:초등과학교육
• /
• 제27권3호
• /
• pp.211-220
• /
• 2008

This study examined pre-service science teachers' perceptions of informal learning by adopting a discourse analysis method suggested by Mortimer and Scott(2003). The guiding research questions were: (1) What are some general patterns of the discourse occurring in science museums between a teacher and a student? (2) In what ways do the pre-service teachers perceive informal learning and teacher's role in informal settings? The 7 pre-service science teachers participated in this study. Each of them shepherd an elementary student around the museum and implemented their own instruction using a pre-planed lesson plan. Results indicated that even though the teachers had learned some characteristics of informal teaming in their college courses, they tended to implement their traditional view of science teaming into the instruction and the view affected them to set up their teaching purposes and contents, and to select communicative approach, patterns of discourse and ways of intervention.

• 한국정보처리학회:학술대회논문집
• /
• 한국정보처리학회 2021년도 춘계학술발표대회
• /
• pp.408-410
• /
• 2021

In the NLP field, the pre-training model BERT launched by the Google team in 2018 has shown amazing results in various tasks in the NLP field. Subsequently, many variant models have been derived based on the original BERT, such as RoBERTa, ERNIEBERT and so on. In this paper, the WWMBERT (Whole Word Masking BERT) model suitable for Chinese text tasks was used as the baseline model of our experiment. The experiment is mainly for "Text-level Chinese text classification tasks" are improved, which mainly combines Tapt (Task-Adaptive Pretraining) and "Multi-Sample Dropout method" to improve the model, and compare the experimental results, experimental data sets and model scoring standards Both are consistent with the official WWMBERT model using Accuracy as the scoring standard. The official WWMBERT model uses the maximum and average values of multiple experimental results as the experimental scores. The development set was 97.70% (97.50%) on the "text-level Chinese text classification task". and 97.70% (97.50%) of the test set. After comparing the results of the experiments in this paper, the development set increased by 0.35% (0.5%) and the test set increased by 0.31% (0.48%). The original baseline model has been significantly improved.