• 소성∙가공
• /
• 제27권2호
• /
• pp.100-106
• /
• 2018

In the continuous casting process, the incidence of transverse surface cracks on the piece may occur by multiple and diverse variables. It is noted that mathematical models may predict only the occurance of the transverse surface cracks, but can require a lot of time (more than three days) to produce a result with this process. This study applied neural networks to predict whether the cracks on the piece surface occurs or does not occur. The computation time was shortened to three minutes, making it applicable to an on-line program, which predicts the non-cracks or cracks of the piece surface in the actual continuous casting process. In addition, the operating conditions to prevent the occurrence of the transverse surface cracks, using decision boundaries were also suggested.

• 한국지능시스템학회논문지
• /
• 제13권2호
• /
• pp.209-215
• /
• 2003

A new GA-based methodology using information granules is suggested for the construction of fuzzy classifiers. The proposed scheme consists of three steps: selection of information granules, construction of the associated fuzzy sets, and tuning of the fuzzy rules. First, the genetic algorithm (GA) is applied to the development of the adequate information granules. The fuzzy sets are then constructed from the analysis of the developed information granules. An interpretable fuzzy classifier is designed by using the constructed fuzzy sets. Finally, the GA are utilized for tuning of the fuzzy rules, which can enhance the classification performance on the misclassified data (e.g., data with the strange pattern or on the boundaries of the classes). To show the effectiveness of the proposed method, an example, the classification of the Iris data, is provided.

• 한국지능시스템학회논문지
• /
• 제13권1호
• /
• pp.107-113
• /
• 2003

A new GA-based methodology using information granules is suggested for the construction of fuzzy classifiers. The proposed scheme consists of three steps: selection of information granules, construction of the associated fuzzy sets, and tuning of the fuzzy rules. First, the genetic algorithm (GA) is applied to the development of the adequate information granules. The fuzzy sets are then constructed from the analysis of the developed information granules. An interpretable fuzzy classifier is designed by using the constructed fuzzy sets. Finally, the GA are utilized for tuning of the fuzzy rules, which can enhance the classification performance on the misclassified data (e.g., data with the strange pattern or on the boundaries of the classes). To show the effectiveness of the proposed method, an example, the classification of the Iris data, is provided.

• 반도체디스플레이기술학회지
• /
• 제22권3호
• /
• pp.161-165
• /
• 2023

One of the difficult works in an autonomous driving system is detecting road lanes or objects in the road boundaries. Detecting and tracking a vehicle is able to play an important role on providing important information in the framework of advanced driver assistance systems such as identifying road traffic conditions and crime situations. This paper proposes a vehicle detection scheme based on deep learning to classify and tracking vehicles in a complex and diverse environment. We use the modified YOLO as the object detector and polynomial regression as object tracker in the driving video. With the experimental results, using YOLO model as deep learning model, it is possible to quickly and accurately perform robust vehicle tracking in various environments, compared to the traditional method.

• 대한한의학원전학회지
• /
• 제27권4호
• /
• pp.181-198
• /
• 2014

Objectives : This study was to analyze definitions of herbal medicinal preparations, crude drug preparations, and new drugs from natural products in the relevant laws and regulations, understand the related problems, and propose directions for improvement. Methods : I analyzed the legal definitions in respect of herbal medicinal preparations, crude drug preparations, and new drugs from natural products in relevant laws and regulations since 1945, explained the problems, and suggested the solution-considering the academic stance of Traditional Korean Medicine and the dualistic medical and pharmaceutical system. Results : Regarding the current laws and regulations that are relevant to herbal medicinal preparations, we should 1) clarify the boundaries between the duty of physicians and that of pharmacists, 2) limit the principles of Korean Medicine as well as the contents of the related textbooks, 3) find a way to protect the intellectual property rights for herbal medicinal preparations, and 4) establish a separate standard for drug classification regarding herbal medicinal preparations. In case of crude drug preparations, we should 1) clarify the meaning and limitations of the phrase, "the point of view of Western medicine," and 2) establish a classification standard for drugs that are used in Korean Medicine and clarify the boundaries between herbal drug preparations and crude drug preparations. Furthermore, laws and regulations apropos of new drugs from natural products do not actually fit the concept of "new drug," and due to subordinate laws, a supplement to a new drug submission is contradictorily misclassified as a new drug from natural products. Conclusions : The problems of legal definitions of herbal medicinal preparations, crude drug preparations, and new drugs from natural products have emerged in the process of giving approval to drugs that are made of herbs and natural products under the dualistic medical and pharmaceutical System. Laws and regulations that differentiate the process of approving herbs that are used in Korean Medicine and the others should be established.

• 대한원격탐사학회:학술대회논문집
• /
• 대한원격탐사학회 2008년도 International Symposium on Remote Sensing
• /
• pp.386-389
• /
• 2008

It is important that the training stage of a supervised classification is designed to provide the spectral information. On the design of the training stage of a classification typically calls for the use of a large sample of randomly selected pure pixels in order to characterize the classes. Such guidance is generally made without regard to the specific nature of the application in-hand, including the classifier to be used. An approach to the training of a support vector machine (SVM) classifier that is the opposite of that generally promoted for training set design is suggested. This approach uses a small sample of mixed spectral responses drawn from purposefully selected locations (geographical boundaries) in training. A sample of such data should, however, be easier and cheaper to acquire than that suggested by traditional approaches. In this research, we evaluated them against traditional approaches with high-resolution satellite data. The results proved that it can be used small mixed pixels to derive a classification with similar accuracy using a large number of pure pixels. The approach can also reduce substantial costs in training data acquisition because the sampling locations used are commonly easy to observe.

• 한국지형학회지
• /
• 제25권3호
• /
• pp.1-18
• /
• 2018

In Korea, 64% of the land is known as mountain area, but the definition and classification standard of mountain are not clear. Demand for utilization and development of mountain area is increasing. In this situation, the unclear definition and scope of the mountain area can lead to the destruction of the mountain and the increase of disasters due to indiscreet permission of forestland use conversion. Therefore, this study analyzed the variables and criteria that can extract the mountain boundaries through the questionnaire survey and the terrain analysis. We developed a mountain boundary extraction algorithm that can classify topographic mountain by using selected variables. As a result, 72.1% of the total land was analyzed as mountain area. For the three catchment areas with different mountain area ratio, we compared the results with the existing data such as forestland map and cadastral map. We confirmed the differences in boundary and distribution of mountain. In a catchment area with predominantly mountainous area, the algorithmbased mountain classification results were judged to be wider than the mountain or forest of the two maps. On the other hand, in the basin where the non-mountainous region predominated, algorithm-based results yielded a lower mountain area ratio than the other two maps. In the two maps, we was able to confirm the distribution of fragmented mountains. However, these areas were classified as non-mountain areas in algorithm-based results. We concluded that this result occurred because of the algorithm, so it is necessary to refine and elaborate the algorithm afterward. Nevertheless, this algorithm can analyze the topographic variables and the optimal value by watershed that can distinguish the mountain area. The results of this study are significant in that the mountain boundaries were extracted considering the characteristics of different mountain topography by region. This study will help establish policies for stable mountain management.

• 한국음향학회지
• /
• 제6권3호
• /
• pp.60-67
• /
• 1987

본 논문에서는 recursive least-square(RLS) 알고리즘을 이용한 한국어 음소분류방법에 관하여 연구하였다. 각 음소의 특징벡터는 prewindowed RLS lattice 알고리즘을 사용하여 추출하는 방법을 제안하였고, 각 음소의 기준패턴은 추출된 특징벡터들을 벡터양자화하여 구성하였다. 제안된 음소인식방식의 성능시험을 위하여 한국어 음소중 자음11개와 모음 8개가 포함된 7개의 한국어 도시명을 발음하여 사용하였으며 초기의 각 음소의 기준패턴으로는 음성신호의 파형을 관찰하여 추출한 표준패턴(prototype)을 사용하였다. 컴퓨터 simulation의 결과로는 화자종속 음소인식의 경우 약간의 음소규칙을 고려할 때 약$85\%$의 음소인식율을 얻었으나, 화자독립 음소인식의 경우는 이보다 훨씬 낮은 인식율을 보였다.

• 한국음향학회지
• /
• 제28권2호
• /
• pp.155-163
• /
• 2009

Text-to-speech 시스템에서 입력 텍스트로부터 운율 정보를 생성하기 위해서는 운율구 경계, 음소 지속시간, 기본주파수 포락선 설정의 3가지 기본적인 모듈이 필요하다. Break 인덱스 (BI; Break Index)는 합성기에서 운율구의 경계를 나타내고, 자연스러운 합성음을 생성하기 위해서는 BI를 정확히 예측하여야 한다. 그러나 BI는 문장의 의미나 화자의 읽기 습관(reading style)에 따라 임의적으로 결정되는 경우가 많아 정확한 예측이 매우 어렵다. 특히 일본어 합성기에서는 악센트 구 경계 (APB; Accentual Phrase Boundary)와 major phrase 경계 (MPB; Major Phrase Boundary)의 정확한 예측이 어렵다. 따라서 본 논문에서는 APB와 MPB 예측 오류를 보완할 수 있는 방법을 제안한다. BI를 고정 break (FB; Fixed Break)와 가변 break (VB; Variable Break)로 분류하여 합성단위 선택을 수행한다. 일반적으로 BI는 한번 생성되면 변하지 않는다. 따라서 BI가 잘못 생성된 경우 최적의 합성음을 생성할 수 없게 되는데, VB는 생성된 BI와 그것과 유사한 BI를 함께 이용하여 합성단위 선택을 수행함으로써 합성음의 BI가 생성된 BI와 다를 수 있는 것을 의미한다. APB와 MPB에 해당하는 BI에 대하여 VB인지 FB인지 CART(Classification and Regression Tree)를 이용하여 예측하고, VB인 경우 기본 주파수와 음소 지속시간에 대해 다중 운율 모델을 생성하여 합성단위 선택을 수행하였다. MOS 테스트 결과 원음이 4.99, 제안한 방법을 4.25, 기존의 방법은 4.01로 합성음의 자연성을 향상시킬 수 있었다.

• 지능정보연구
• /
• 제14권2호
• /
• pp.193-206
• /
• 2008

의료 진단 문제는 기정의된 특성치들로 표현되는 환자의 상태 데이터로부터 병의 유무를 판단하는 일종의 분류 문제로 간주할 수 있다. 본 연구는 혼용 유전자 알고리즘 기반의 분류방법을 도입함으로써 의료 진단 문제와 같은 다차원의 패턴 분류 문제를 해결할 수 있는 방안을 제안하고 있다. 일반적으로 분류 문제는 데이터 패턴에 존재하는 여러 클래스 간 구분경계를 생성하는 접근방법을 사용하는데, 이를 위해 본 연구에서는 일단의 영역 에이전트들을 도입하여 이들을 유전자 알고리즘 및 국소 적응조작을 혼용함으로써 데이터 패턴에 적응하도록 유도하고 있다. 일반적인 유전자 알고리즘의 진화단계를 거친 에이전트들에 적용되는 국소 적응조작은 영역 에이전트의 확장, 회피 및 재배치로 이루어지며, 각 에이전트의 적합도에 따라 이들 중 하나가 선택되어 해당 에이전트에 적용된다. 제안된 의료 진단용 분류 방법은 UCI 데이터베이스에 있는 잘 알려진 의료 데이터, 즉 간, 당뇨, 유방암 관련 진단 문제에 적용하여 검증하였다. 그 결과, 기존의 대표적인 분류기법인 최단거리이웃방법(the nearest neighbor), C4.5 알고리즘에 의한 의사 결정트리(decision tree) 및 신경망보다 우수한 진단 수행도를 나타내었다.