• 한국컴퓨터정보학회논문지
• /
• 제24권1호
• /
• pp.33-39
• /
• 2019

In this paper, we propose deep convolutional neural network(CNN) with bottleneck structure which improves the performance of earthquake classification. In order to address all possible forms of earthquakes including micro-earthquakes and artificial-earthquakes as well as large earthquakes, we need a representation and classifier that can effectively discriminate seismic waveforms in adverse conditions. In particular, to robustly classify seismic waveforms even in low snr, a deep CNN with 1x1 convolution bottleneck structure is proposed in raw seismic waveforms. The representative experimental results show that the proposed method is effective for noisy seismic waveforms and outperforms the previous state-of-the art methods on domestic earthquake database.

• 한국멀티미디어학회논문지
• /
• 제23권7호
• /
• pp.803-811
• /
• 2020

This paper proposes an image classification algorithm that transforms the number of convolution layers in the residual block of ResNet, CNN's representative method. The proposed method modified the structure of 34/50 layer of ResNet structure. First, we analyzed the performance of small and many convolution layers for the structure consisting of only shortcut and 3 × 3 convolution layers for 34 and 50 layers. And then the performance was analyzed in the case of small and many cases of convolutional layers for the bottleneck structure of 50 layers. By applying the results, the best classification method in the residual block was applied to construct a 34-layer simple structure and a 50-layer bottleneck image classification model. To evaluate the performance of the proposed image classification model, the results were analyzed by applying to the cifar10 dataset. The proposed 34-layer simple structure and 50-layer bottleneck showed improved performance over the ResNet-110 and Densnet-40 models.

• Asian-Australasian Journal of Animal Sciences
• /
• 제18권7호
• /
• pp.915-921
• /
• 2005

Genetic variation at 25 microsatellite loci, population structure, and genetic bottleneck hypothesis were examined for Ankleshwar poultry population found in Gujrat, India. The estimates of genetic variability such as effective number of alleles and gene diversities revealed substantial genetic variation frequently displayed by microsatellite markers. The average polymorphism across the studied loci and the expected gene diversity in the population were 6.44 and 0.670${\pm}$0.144, respectively. The population was observed to be significantly differentiated into different groups, and showed fairly high level of inbreeding (f = 0.240${\pm}$0.052) and global heterozygote deficit. The bottleneck analysis indicated the absence of genetic bottleneck in the past. The study revealed that the Ankleshwar poultry breed needs appropriate genetic management for its conservation and improvement. The information generated in this study may further be utilized for studying differentiation and relationships among different Indian poultry breeds.

• 대한수학회지
• /
• 제35권3호
• /
• pp.727-740
• /
• 1998

Controlling traffic lights at a bottleneck, in [5] a time of open passage is called optimal, if it minimizes the first moment of the asymptotic distribution of the queue length. The discussion of the first moment as function of the time of open passage is based on an analysis of the behavior of a fixed point when varying control parameters and delivers theoretical and computational aspects of the traffic problem.

• 대한전자공학회:학술대회논문집
• /
• 대한전자공학회 2002년도 ITC-CSCC -1
• /
• pp.66-69
• /
• 2002

This paper proposes a new dynamic storage structure and methods fur geometric data with detail levels. Using geometric data with detail levels, we can search geometric data quickly. However, the previous structures for detail levels form the bottleneck in the design of database and do not support all types of geometric data with detail levels. Our structure supports all types of geometric data with detail levels. Moreover, our structure does not form bottleneck in the design of database. This paper presents the structure and algorithms for searching and updating of geometric data with detail levels. Experiments are then performed.

• Journal of Information Processing Systems
• /
• 제17권5호
• /
• pp.892-904
• /
• 2021

An image semantic segmentation model is proposed based on improved ENet network in order to achieve the low accuracy of image semantic segmentation in complex environment. Firstly, this paper performs pruning and convolution optimization operations on the ENet network. That is, the network structure is reasonably adjusted for better results in image segmentation by reducing the convolution operation in the decoder and proposing the bottleneck convolution structure. Squeeze-and-excitation (SE) module is then integrated into the optimized ENet network. Small-scale targets see improvement in segmentation accuracy via automatic learning of the importance of each feature channel. Finally, the experiment was verified on the public dataset. This method outperforms the existing comparison methods in mean pixel accuracy (MPA) and mean intersection over union (MIOU) values. And in a short running time, the accuracy of the segmentation and the efficiency of the operation are guaranteed.

• 전기전자학회논문지
• /
• 제25권3호
• /
• pp.493-500
• /
• 2021

본 논문에서는 다각형 용기의 품질 향상을 위한 딥러닝 구조 개발을 제안한다. 딥러닝 구조는 convolution 층, bottleneck 층, fully connect 층, softmax 층 등으로 구성된다. Convolution 층은 입력 이미지 또는 이전 층의 특징 이미지를 여러 특징 필터와 convolution 3x3 연산하여 특징 이미지를 얻어 내는 층이다. Bottleneck 층은 convolution 층을 통해 추출된 특징 이미지상의 특징들 중에서 최적의 특징들만 선별하여 convolution 1x1 ReLU로 채널을 감소시키고convolution 3x3 ReLU를 실시한다. Bottleneck 층을 거친 후에 수행되는 global average pooling 연산과정은 convolution 층을 통해 추출된 특징 이미지의 특징들 중에서 최적의 특징들만 선별하여 특징 이미지의 크기를 감소시킨다. Fully connect 층은 6개의 fully connect layer를 거쳐 출력 데이터가 산출된다. Softmax 층은 입력층 노드의 값과 연산을 진행하려는 목표 노드 사이의 가중치와 곱을 하여 합하고 활성화 함수를 통해 0~1 사이의 값으로 변환한다. 학습이 완료된 후에 인식 과정에서는 학습 과정과 마찬가지로 카메라를 이용한 이미지 획득, 측정 위치 검출, 딥러닝을 활용한 비원형 유리병 분류 등을 수행하여 비원형 유리병을 분류한다. 제안된 다각형 용기의 품질 향상을 위한 딥러닝 구조의 성능을 평가하기 위하여 공인시험기관에서 실험한 결과, 양품/불량 판별 정확도 99%로 세계최고 수준과 동일한 수준으로 산출되었다. 검사 소요 시간은 평균 1.7초로 비원형 머신비전 시스템을 사용하는 생산 공정의 가동 시간 기준 내로 산출되었다. 따라서 본 본문에서 제안한 다각형 용기의 품질 향상을 위한 딥러닝 구조의 성능의 그 효용성이 입증되었다.

• 한국정보과학회논문지:정보통신
• /
• 제29권1호
• /
• pp.40-47
• /
• 2002

본 논문은 작업분석(Operational Analysis)을 이용하여 ATM 기반 수동 광 가입자 망(Passive Optical Network: PON)에서 광선로 종단 시스템의 성능을 분석하였다. 이를 위해 ATM-PON을 구성하는 각 블록에서 버퍼 구조에 따른 이용율, 쓰루풋, 그리고 라우팅 속도와 같은 시스템 파라미터를 사용하였다. 특히, 각 블록의 평균 응답시간과 방문률(visit ratio)를 구하고 시스템 성능을 저해하는 병목점을 찾았다. 본 논문에서는 622Mb/s 16x16 스위칭 블록이 병목점 블록임을 확인하였으며, 이 병목점은 셀 도착율이 증가함에 따라 손실률이 급격하게 증가함에 확인하였다.

• 디지털융복합연구
• /
• 제6권1호
• /
• pp.43-52
• /
• 2008

Managing of local government public service business is innovation strategy of local government public service business using choice and competition principle. Managing strategy of local government public service business is new paradigm for innovating public service business and role structure of government. It is important to understand properly substance and characteristics of local government public service business. Accordingly, the purpose of this study find bottleneck of local government public service business and take out improvement plan and suggest policy plan of Seo Gu District Office in Gwang Ju City in the future.

• 전자공학회논문지S
• /
• 제34S권11호
• /
• pp.102-113
• /
• 1997

In this paepr, a new image segmentation method reducing efficiently contour information causing bottleneck problem at segmentatio-based very low bit rate codingis proposed, while preserving objective and subjective quality. It consists of 4-level hierarchical image segmentation based on mathematical morphology and 1-leve region merging structure using contast of two adjacent regions. For two adjacent region pairs at the fourth level included in each region of the thid level, contrast is calculated. Among the pairs of two adjacent regions with less value than threshold, two adjacent regions having the minimum contrast are merged first. After region merging, texture of the merged region is updated. The procedure is performed recursively for all the adjacent region pairs at the fourth level included in each region of the third level. Compared with the previous method, the objective and subjective image qualities are similar. But it reduces 46.65% texture information on the average by decreasing total region number to be tansmitted. Specially, it shows reduction of the 23.95% contour information of the average. Thus, it can improve efficiently the bottleneck problem at segementation-based very low bit rate coding.