• 제어로봇시스템학회:학술대회논문집
• /
• 제어로봇시스템학회 1990년도 한국자동제어학술회의논문집(국제학술편); KOEX, Seoul; 26-27 Oct. 1990
• /
• pp.951-956
• /
• 1990

The present paper describes a new technique for associating images employing a set of local constraints among pixels on an image. The technique describes the association problem in terms of consistent labeling which is an abstraction of various kinds of network constraints problems. In this particular research, a pixel and its gray value correspond to a unit and a label, respectively. Since constraints among units on an image are defined with respect to each n-tuple of pixels, performance of the present association technique largely depends on how to choose the n-tuples on an image plane. The main part of this paper is devoted to discussing this selection scheme and giving a solution to it as well as showing the algorithm of association. Also given are some results of the simulation performed on synthetic binary images to examine the performance of proposed technique, followed by the argument on further studies.

• 영재교육연구
• /
• 제5권2호
• /
• pp.17-36
• /
• 1995

The present study examined preschooler's (3-5yrs) representation and evaluation skills in a puzzle completion task. The puzzle contained panels of four children dressed for each seacon and the key to success was using a body scheme to reconstruct the panels (head, torso, legs, feet and sky on top). Baseline data (Study 1) revealed a developmental pattern of increasing bydy scheme representation along with more careful attention to season consitent construction. Spontaneous verbalization also shifted from more guiding statements (where'the head?) to move evaluative statements (this isn't right). Study 2 examined different intervention techniques for increasing representation (verbal laveling) and evaluative processes (error detection practice), along with a control group that had unassisted practice. Three year olds benefited from verbal labeling, four year olds from both types of training. Verbalizations also showed appropriated shifts toward increasing evaluation, particularly for the older children. These findings are discussed in terms of a developmental hypothesis that representation precedes evaluation skills and that training techniques should take into account the relative balance between representation and evaluation skills in the individual for the task at hand.

• Macromolecular Research
• /
• 제17권6호
• /
• pp.403-410
• /
• 2009

Cytotoxicity is a severe problem of cadmium sulfide nanoparticles(CSNPs) for use in biological systems. In the present study, mercaptoacetic acid-coated CSNPs were conjugated with bovine serum albumin (BSA) to improve biocompatibility. The surface properties of the CSNPs and albumin-conjugated CSNPs (ACSNPs) were characterized by XRD, UV, FTIR, EA, TEM and DLS. Human breast cancer cells (KB cells) were then cultured in the presence of the nanoparticles to evaluate the cytotoxicity of CSNPs and ACSNPs. Finally, the fluorescence intensity of the nanoparticles' aqueous solution was examined using a fluorescence spectrometer. The results showed that the cell compatibility and fluorescence intensity of ACSNPs were higher than those of CSNPs. The strongly luminescent features of the biocompatible ACSNPs are promising for use in biological fields such as cellular labeling, intracellular tracking and molecular imaging.

• 한국CDE학회논문집
• /
• 제19권4호
• /
• pp.402-409
• /
• 2014

Presented in this paper is a scheduling method for semiconductor backend process considering the backward pegging. It is known that the pegging for frontend is a process of labeling WIP lots for target order which is specified by due date, quantity, and product specifications including customer information. As a result, it gives the release plan to meet the out target considering current WIP. However, the semiconductor backend process includes the multichip package and test operation for the product bin portion. Therefore, backward pegging method for frontend can't give the release plan for backend process in semiconductor. In this paper, we suggest backward pegging method considering the characteristics of multichip package and test operation in backend process. And we describe the backward pegging problem using the examples.

• ETRI Journal
• /
• 제39권5호
• /
• pp.652-661
• /
• 2017

A mention has a noun or noun phrase as its head and constructs a chunk that defines any meaning, including a modifier. Mention detection refers to the extraction of mentions from a document. In mentions, coreference resolution refers to determining any mentions that have the same meaning. Pointer networks, which are models based on a recurrent neural network encoder-decoder, outputs a list of elements corresponding to an input sequence. In this paper, we propose mention detection using pointer networks. This approach can solve the problem of overlapped mention detection, which cannot be solved by a sequence labeling approach. The experimental results show that the performance of the proposed mention detection approach is F1 of 80.75%, which is 8% higher than rule-based mention detection, and the performance of the coreference resolution has a CoNLL F1 of 56.67% (mention boundary), which is 7.68% higher than coreference resolution using rule-based mention detection.

• 한국정보과학회 언어공학연구회:학술대회논문집(한글 및 한국어 정보처리)
• /
• 한국정보과학회언어공학연구회 2015년도 제27회 한글 및 한국어 정보처리 학술대회
• /
• pp.194-197
• /
• 2015

Long Short-term Memory Network(LSTM) 기반 Recurrent Neural Network(RNN)는 순차 데이터를 모델링 할 수 있는 딥 러닝 모델이다. 기존 RNN의 그래디언트 소멸 문제(vanishing gradient problem)를 해결한 LSTM RNN은 멀리 떨어져 있는 이전의 입력 정보를 볼 수 있다는 장점이 있어 음성 인식 및 필기체 인식 등의 분야에서 좋은 성능을 보이고 있다. 또한 LSTM RNN 모델에 의존성(전이 확률)을 추가한 LSTM CRF모델이 자연어처리의 한 분야인 개체명 인식에서 우수한 성능을 보이고 있다. 본 논문에서는 한국어 문장의 지배소가 문장 후위에 나타나는 점에 착안하여 Backward 방식의 LSTM CRF 모델을 제안하고 이를 한국어 의미역 결정에 적용하여 기존 연구보다 더 높은 성능을 얻을 수 있음을 보인다.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제9권10호
• /
• pp.4092-4107
• /
• 2015

We propose a background subtraction method for moving cameras based on trajectory classification, image segmentation and label inference. In the trajectory classification process, PCA-based outlier detection strategy is used to remove the outliers in the foreground trajectories. Combining optical flow trajectory with watershed algorithm, we propose a trajectory-controlled watershed segmentation algorithm which effectively improves the edge-preserving performance and prevents the over-smooth problem. Finally, label inference based on Markov Random field is conducted for labeling the unlabeled pixels. Experimental results on the motionseg database demonstrate the promising performance of the proposed approach compared with other competing methods.

• 융복합기술연구소 논문집
• /
• 제11권1호
• /
• pp.1-5
• /
• 2021

Deep learning based methods achieve state-of-the-art accuracy, however, they typically rely on supervised training with large labeled datasets. It is known in many medical applications that labeling medical images requires significant expertise and much time, and typical hand-tuned approaches for data augmentation fail to capture the complex variations in such images. This paper proposes a 3D image augmentation method to overcome these difficulties. It allows us to enrich diversity of training data samples that is essential in medical image segmentation tasks, thus reducing the data overfitting problem caused by the fact the scale of medical image dataset is typically smaller. Our numerical experiments demonstrate that the proposed approach provides significant improvements over state-of-the-art methods for 3D medical image segmentation.

• 한국정보과학회 언어공학연구회:학술대회논문집(한글 및 한국어 정보처리)
• /
• 한국정보과학회언어공학연구회 2022년도 제34회 한글 및 한국어 정보처리 학술대회
• /
• pp.35-40
• /
• 2022

본 논문에서는 시퀀스 레이블링 문제(sequence labeling problem)인 개체명 인식에 사용할 새로운 태깅 포맷인 Delimiter tag (D-tag)를 소개한다. 시퀀스 레이블링 문제에서 사용하는 BIO-tag 포맷은 개체명 레이블을 B (beginning)와 I (inside) 의미의 레이블로 확장하여 타겟 클래스의 수가 2배 증가한다. 또한 BIO-tag 포맷을 사용할 경우, 모델이 B와 I 를 잘못 분류하는 문제가 발생하며, 레이블 수가 많은 세부 분류 개체명의 경우에는 label confusion을 야기한다. 본 논문에서 제안한 D-tag 포맷은 타겟 클래스의 수를 증가시키지 않기 때문에 앞서 언급한 문제를 해결할 수 있다. 실험 결과, D-tag를 사용하여 학습한 모델이 BIO-tag를 사용한 경우보다 더 좋은 성능을 보여, 유망함을 확인하였다.

• 한국정보통신학회:학술대회논문집
• /
• 한국정보통신학회 2021년도 추계학술대회
• /
• pp.100-101
• /
• 2021

고령 사회에 들어선 한국은 노인 인구의 87%가 치매, 중풍 등 만성질환을 앓고 있으며 이중 알츠하이머 치매는 전체 치매의 71.3%를 차지할 정도로 치매 중 높은 비율로 나타난다. 본 논문은 알츠하이머 치매 MRI 이미지를 3단계로 나눈 딥러닝 결과의 오차 문제를 검토하기 위해 라벨링 검증을 하였다.