• Journal of the Ergonomics Society of Korea
• /
• v.30 no.1
• /
• pp.9-17
• /
• 2011

Consumer products are produced on the premise that consumers can use their products safely and effectively no matter how serious human errors they may make. However, different careers and educational experiences of them may induce diverse human errors when they want to use them. In that sense, not a few policies to reduce human errors may show some implications for human error prevention and industrial design of consumer products. In this paper, producers' safety efforts required by Product Liability(PL) Act were reviewed in view of human error prevention, and legal aspects of manufacturers' responsibility for consumer products were discussed in relation to Product Liability Act. Then, principal approaches for them were introduced under the title of System Safety Precedence. After that, major key points for preventing human errors related with consumer products - such as ergonomic design and effective labeling - were discussed with reference to ISO standards. Therefore, it was shown that all the efforts required by PL Act would be correspondent to human error prevention in the whole manufacturing processes if understood by ergonomists. To make a conclusion, it could be said that, for human error prevention, the principle of System Safety Precedence would be indispensable, and that all the efforts for preventing human errors should be systematically organized in Product Safety Management Systems.

• 한국어정보학회:학술대회논문집
• /
• 2017.10a
• /
• pp.56-59
• /
• 2017

CNN(Convolutional Neural Network)을 이용하여 발화 주제 다중 분류 task를 multi-labeling 방법과, cluster 방법을 이용하여 수행하고, 각 방법론에 MSE(Mean Square Error), softmax cross-entropy, sigmoid cross-entropy를 적용하여 성능을 평가하였다. Network는 음절 단위로 tokenize하고, 품사정보를 각 token의 추가한 sequence와, Naver DB를 통하여 얻은 named entity 정보를 입력으로 사용한다. 실험결과 cluster 방법으로 문제를 변형하고, sigmoid를 output layer의 activation function으로 사용하고 cross entropy cost function을 이용하여 network를 학습시켰을 때 F1 0.9873으로 가장 좋은 성능을 보였다.

• Journal of Internet Computing and Services
• /
• v.18 no.1
• /
• pp.65-75
• /
• 2017

The Detection Model is the model to find the result of a certain purpose using artificial intelligent, data mining, intelligent algorithms In Cyber Security, it usually uses to detect intrusion, malwares, cyber incident, and attacks etc. There are an amount of unlabeled data that are collected in a real environment such as security data. Since the most of data are not defined the class labels, it is difficult to know type of data. Therefore, the label determination process is required to detect and analysis with accuracy. In this paper, we proposed a KDFL(K-means and D-S Fusion based Labeling) method using D-S inference and k-means(unsupervised) algorithms to decide label of data records by fusion, and a detection model architecture using a proposed labeling method. A proposed method has shown better performance on detection rate, accuracy, F1-measure index than other methods. In addition, since it has shown the improved results in error rate, we have verified good performance of our proposed method.

• Journal of Radiopharmaceuticals and Molecular Probes
• /
• v.3 no.2
• /
• pp.54-58
• /
• 2017

Matrix-assisted laser desorption/ionization time-of-flight mass spectrometry(MALDI-TOF MS) is one of the powerful methods that enable analysis of small molecules as well as large molecules up to about 500,000 Da without severe fragmentation. MALDI-TOF MS, thus, has been a very useful an analytical tool for the confirmation of synthetic molecules, probing PTMs, and identifying structures of a given protein. In recent nuclear medicine, MALDI-TOF MS liner ion mode helps researcher calculate the average number of chelator(or linkage) per an antibody conjugate, such as DOTA-(or DFO-) trastuzumab for labeling a medical radioisotope. This simple technique can be utilized to improve the labeling method and control the quality at the development of antibody-based radiopharmaceuticals, which is very effected to diagnosis and therapy for in vivo tumor cells, with radioisotopes like $^{89}Zr$, $^{64}Cu$, and 177Lu. To minimize the error, MALDI-TOF MS measurement is repeatedly performed for each sample in this study, and external calibration is carried out after data collection.

• The Journal of The Korea Institute of Intelligent Transport Systems
• /
• v.13 no.5
• /
• pp.35-49
• /
• 2014

This paper proposes the on-road object detection and classification algorithm by using a detection system consisting of only laser scanners. Each sensor data acquired by the laser scanner is fused with a grid map and the measurement error and spot spaces are corrected using a labeling method and dilation operation. Fuzzy method which uses the object information (length, width) as input parameters can classify the objects such as a pedestrian, bicycle and vehicle. In this way, the accuracy of the detection system is increased. Through experiments for some scenarios in the real road environment, the performance of the proposed detection and classification system for the actual objects is demonstrated through the comparison with the actual information acquired by GPS-RTK.

• Transactions of the Korean Society of Mechanical Engineers A
• /
• v.25 no.5
• /
• pp.841-849
• /
• 2001

The pixel information of the object was obtained sequentially and pixels were clustered to a label by the line labeling method. Feature points were determined by finding the slope for edge pixels after selecting the fixed number of edge pixels. The slope was estimated by the least square method to reduce the detection error. Once a matching point was determined by comparing the feature information of the object and the pattern, the parameters for translation, scaling and rotation were obtained by selecting the longer line of the two which passed through the matching point from left and right sides. Finally, modified Hausdorff Distance has been used to identify the similarity between the object and the given pattern. The multi-label method was developed for recognizing the patterns with more than one label, which performs the modified Hausdorff Distance twice. Experiments have been performed to verify the performance of the proposed algorithm and method for simple target image, complex target image, simple pattern, and complex pattern as well as the partially hidden object. It was proved via experiments that the proposed image matching algorithm for recognizing the object had a good performance of matching.

• ETRI Journal
• /
• v.37 no.3
• /
• pp.541-550
• /
• 2015

In this paper, we propose a classification-based approach for hybridizing statistical machine translation and rulebased machine translation. Both the training dataset used in the learning of our proposed classifier and our feature extraction method affect the hybridization quality. To create one such training dataset, a previous approach used auto-evaluation metrics to determine from a set of component machine translation (MT) systems which gave the more accurate translation (by a comparative method). Once this had been determined, the most accurate translation was then labelled in such a way so as to indicate the MT system from which it came. In this previous approach, when the metric evaluation scores were low, there existed a high level of uncertainty as to which of the component MT systems was actually producing the better translation. To relax such uncertainty or error in classification, we propose an alternative approach to such labeling; that is, a cut-off method. In our experiments, using the aforementioned cut-off method in our proposed classifier, we managed to achieve a translation accuracy of 81.5% - a 5.0% improvement over existing methods.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.9 no.6
• /
• pp.2302-2316
• /
• 2015

We propose an efficient method of extracting targets within a region of interest in non-homogeneous infrared images by using a principal component analysis (PCA) plane and adaptive Gaussian kernel. Existing approaches for extracting targets have been limited to using only the intensity values of the pixels in a target region. However, it is difficult to extract the target regions effectively because the intensity values of the target region are mixed with the background intensity values. To overcome this problem, we propose a novel PCA based approach consisting of three steps. In the first step, we apply a PCA technique minimizing the total least-square errors of an IR image. In the second step, we generate a binary image that consists of pixels with higher values than the plane, and then calculate the second derivative of the sum of the square errors (SDSSE). In the final step, an iteration is performed until the convergence criteria is met, including the SDSSE, angle and labeling value. Therefore, a Gaussian kernel is weighted in addition to the PCA plane with the non-removed data from the previous step. Experimental results show that the proposed method achieves better segmentation performance than the existing method.

• ETRI Journal
• /
• v.27 no.3
• /
• pp.267-272
• /
• 2005

The transmission performance of optical labeling based on a combined frequency shift keying/amplitude shift keying (FSK/ASK) format is studied by numerical simulation. The simulation demonstrates that the bit-error ratio (BER) characteristic of an ASK signal is limited by the extinction ratio, received optical power, and dispersion, simultaneously. However, an FSK signal is mainly limited by the extinction ratio (ER) and received optical power when the peak spectrum, which is used to detect the FSK signal, is relatively narrow.

• Annual Conference on Human and Language Technology
• /
• 2017.10a
• /
• pp.203-208
• /
• 2017

본 논문은 한글 자모 조합 임베딩을 이용하여 오타에 강건한 한국어 품사 태깅 시스템을 구축하는 방법에 대해 기술한다. 최근 딥 러닝 연구가 활발히 진행되면서 자질을 직접 추출해야 하는 기존의 기계학습 방법이 아닌, 스스로 자질을 찾아서 학습하는 딥 러닝 모델을 이용한 연구가 늘어나고 있다. 본 논문에서는 다양한 딥 러닝 모델 중에서 sequence labeling에 강점을 갖고 있는 bidirectional LSTM CRFs 모델을 사용하였다. 한국어 품사 태깅 문제에서 일반적으로 사용되는 음절 임베딩은 약간의 오타에도 품사 태깅 성능이 크게 하락하는 한계가 있었다. 따라서 이를 개선하기 위해 본 논문에서는 한글 자모 임베딩 값을 조합시킨 음절 임베딩 방식을 제안하였다. 강제로 오타를 발생시킨 테스트 집합에서 실험한 결과, 자모 조합 임베딩 기법이 word2vec 음절 임베딩 방식에 비해 형태소 분할은 0.9%, 품사 태깅은 3.5% 우수한 성능을 기록하였다.