• Korean Chemical Engineering Research
• /
• v.59 no.2
• /
• pp.186-190
• /
• 2021

The toxics regulatory body provides a benefit/cost ratio as a justification criterion while implementing regulations that induce the industry to reduce emissions voluntarily. Furthermore, since, the body wants to reflect not only the efficiency standard but also the policy standard in the evaluation of feasibility, it calculates the ratio by adjusting the importance weights. The problem is that respondents answer ambiguously. It should be removed for the reasonable evaluation. This study introduced a fuzzy-AHP methodology for this, and applied it to the voluntary emission reduction plan program in Korea.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2020.07a
• /
• pp.662-663
• /
• 2020

최근 MPEG 에서는 인공신경망 모델을 다양한 딥러닝 프레임워크에서 상호운용 가능한 포맷으로 압축 표현할 수 있는 NNR(Compression of Neural Network for Multimedia Content Description and Analysis) 표준화를 진행하고 있다. 본 논문에서는 MPEG-NNR 에서 CNN 모델을 압축하기 위한 지역 비선형 양자화(Local Non-linear Quantization: LNQ) 기법을 제시한다. 제안하는 LNQ 는 균일 양자화된 CNN 모델의 각 계층의 가중치 행렬 블록 단위로 추가적인 비선형 양자화를 적용한다. 또한, 제안된 LNQ 는 가지치기(pruning)된 모델의 경우 블록내의 영(zero) 값의 가중치들은 그대로 전송하고 영이 아닌 가중치만을 이진 군집화를 적용한다. 제안 기법은 음성 분류를 위한 CNN 모델(DCASE Task)의 압축 실험에서 기존 균일 양자화를 대비 동일한 분류 성능에서 약 1.78 배 압축 성능 향상이 있음을 확인하였다.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2024.05a
• /
• pp.814-817
• /
• 2024

특허의 가치를 평가하기 위해서는 특허 데이터에 포함된 다양한 지표가 활용될 수 있으며, 최근 다양한 지표를 머신 러닝 기법으로 분석하여 특허의 가치를 평가하는 연구가 증가하고 있다. 특허의 가치를 올바르게 평가하기 위해서는 여러 지표 중에서 어떤 지표가 특허의 가치에 크게 기여 하는지 판단할 수 있어야 하며, 이에 따라 지표별로 적절한 가중치를 설정할 수 있어야 한다. 제안된 방법은 회귀 모델 기반으로 다양한 지표에 가중치를 적용하여 특허 피인용수를 예측하였으며, 특허 패밀리 수에 적용되는 가중치를 변경하면서 특허 패밀리 수가 특허의 가치에 미치는 영향을 검증하였고, 특허 가치 평가 과정에서 특허 패밀리 수의 중요성에 대해 확인하였다.

• Journal of Digital Convergence
• /
• v.11 no.12
• /
• pp.201-211
• /
• 2013

This study compares Customer Satisfaction Index(CSI) and the weight for each dimension by applying various methods of weight calculation and attempts to suggest some implications. For the purpose, the study classified the methods of weight calculation into the subjective method and the statistical method. Constant sum scale was used for the subjective method, and the statistical method was again segmented into correlation analysis, principal component analysis, factor analysis, structural equation model. The findings showed that there is difference between the weights from the subjective method and the statistical method. The order of the weights by the analysis methods were classified with similar patterns. Besides, the weight for each dimension by different methods of weight calculation showed considerable deviation and revealed the difference of discrimination and stability among the dimensions. Lastly, the CSI calculated by various methods of weight calculation showed to be the highest in structural equation model, followed by in the order of regression analysis, correlation analysis, arithmetic mean, principal component analysis, constant sum scale and factor analysis. The CSI calculated by each method showed to have statistically significant difference.

• Journal of Broadcast Engineering
• /
• v.21 no.6
• /
• pp.889-898
• /
• 2016

Virtual Bass System (VBS) is widely used to extend the lower frequency limit of small loudspeakers, which generates harmonics of a fundamental frequency. The perceptual quality of the VBS is highly dependent on the harmonic weighting strategy. There have been several weighting methods, including exponential attenuation and timbre matching. However, it is essential to match phases between harmonics in the original signal and generate harmonics to precisely convey the weighting strategy. This paper shows the limitations of the previous harmonic weighting schemes and proposes a new harmonic weighting scheme. The proposed weighting scheme proposes phase matching between the original and generated harmonics and varies the slope of the attenuation weighting dynamically according to the missing fundamental frequency. Objective and subjective tests show that the proposed harmonic weighting scheme provides more natural and effective bass perception in a limited situation than the conventional schemes, which implies that the phase matching is essential for the high quality bass enhancement.

• Journal of the Institute of Electronics and Information Engineers
• /
• v.50 no.5
• /
• pp.208-217
• /
• 2013

When a beamforming can be processed separately in horizontal and vertical directions with the planar arrays used in sonar systems, there are several merits such as that practically reduce the required computations and volumes. However, the common planar arrays used in sonar systems are generally non-separable, so the beamforming with separable weighting results in the differences between the desired beam characteristics and the resultant beam characteristics. In this paper, we propose a new separable weighting method which can achieve the wanted beam characteristics by using the separable weights in horizontal and vertical directions for the non-separable planar arrays. In order to achieve the wanted beam characteristics, the proposed method minimizes the differences between the desired weights and the resultant weights based on the number of effective sensors in horizontal and vertical directions of the planar arrays.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.26 no.7A
• /
• pp.1152-1161
• /
• 2001

본 논문에서는 이동국 수신 신호의 SINR을 최대화하는 빔 성형 알고리듬을 이용하여, 폐쇄 루프 송신 빔 성형기술을 적용한 순방향 링크 CDMA 시스템에서의 안테나 수, 귀환 채널의 구조, 귀환 지연 등과 시스템 성능과의 상관관계를 연구하였다. 폐쇄 루프 전송 빔 성형에서는 이동국이 각 안테나가 겪는 채널을 추정하여 수신 SINR을 최대화시키는 가중치 벡터를 계산하고, 귀환채널을 통해 가중치 벡터의 양자화된 진폭과 위상정보를 전송한다. 컴퓨터 모의 실험 결과는 송신 안테나가 2개, 3개, 4개로 늘어남에 따라 빔 성형 이득은 단일 송신 안테나와 비교해 $10^{-5}$ BER 근방에서 4.2dB, 5.8dB, 7dB로 증가하지만 양자화 오류에 의한 성능 저하 또한 0.1dB, 0.6dB, 1.3dB로 커지는 것을 보여준다. 또한 순방향 채널의 최대 도플러 주파수가 100Hz일 때에는 귀환 채널을 통한 소신 가중치 벡터를 보다 빠르게 갱신하는 것이 가중치 벡터의 양자화 레벨의 수를 늘여주는 것보다 $10^{-5}$ BER 근방에서 0.6dB 더 좋은 성능을 보이며, 최대 도플러 주파수가 10Hz일 때에는 가중치 벡터의 갱신 속도를 늘이기 보다 양자화 레벨의 수를 늘여주어 정확한 가중치 벡터를 전송하는 편이 0.9dB의 성능 향상을 보인다. 두 전력제어 그룹 길이의 귀환지연으로 인한 성능저하는 채널의 최대 도플러 주파수가 50Hz인 경우가 채널의 최대 도플러 주파수가 100Hz인 경우보다 $10^{-5}$ BER 근방에서 0.3dB 정도 더 작다. 또한 AOS가 3$^{\circ}$인 경우가 AOS가 $10^{\circ}$인 경우보다 $10^{-5}$ BER 근방에서 1.9dB 정도, 주파수 선택적 페이딩 채널이 주파수 비선택적 페이딩 채널보다 $10^{-5}$ BER 근방에서 1dB 정도 귀환 지연으로 인한 성능의 저하가 작다.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.18 no.3
• /
• pp.706-712
• /
• 2014

This paper describes a determining method of weighting factors for two-point interpolation filters. The interpolation filters is implemented by applying modifying functions to the linear interpolation. Here, there is a problem of determining weights when modifying functions being engaged. The previous method determined the weights by imposing c1-continuity on the interpolation kernels. However, this approach is unable to use the property of individual modifying functions. In this paper, on the basis of spectral analyses of the modifying functions and image signals, we provide a determination method by experimental results. Thus, many experiments are carried out to do so. The results indicate that different weights are required for different modifying functions and also the proposed method outperforms than the previous method.

• Journal of Intelligence and Information Systems
• /
• v.21 no.1
• /
• pp.29-45
• /
• 2015

Response modeling is a well-known research issue for those who have tried to get more superior performance in the capability of predicting the customers' response for the marketing promotion. The response model for customers would reduce the marketing cost by identifying prospective customers from very large customer database and predicting the purchasing intention of the selected customers while the promotion which is derived from an undifferentiated marketing strategy results in unnecessary cost. In addition, the big data environment has accelerated developing the response model with data mining techniques such as CBR, neural networks and support vector machines. And CBR is one of the most major tools in business because it is known as simple and robust to apply to the response model. However, CBR is an attractive data mining technique for data mining applications in business even though it hasn't shown high performance compared to other machine learning techniques. Thus many studies have tried to improve CBR and utilized in business data mining with the enhanced algorithms or the support of other techniques such as genetic algorithm, decision tree and AHP (Analytic Process Hierarchy). Ahn and Kim(2008) utilized logit, neural networks, CBR to predict that which customers would purchase the items promoted by marketing department and tried to optimized the number of k for k-nearest neighbor with genetic algorithm for the purpose of improving the performance of the integrated model. Hong and Park(2009) noted that the integrated approach with CBR for logit, neural networks, and Support Vector Machine (SVM) showed more improved prediction ability for response of customers to marketing promotion than each data mining models such as logit, neural networks, and SVM. This paper presented an approach to predict customers' response of marketing promotion with Case Based Reasoning. The proposed model was developed by applying different weights to each feature. We deployed logit model with a database including the promotion and the purchasing data of bath soap. After that, the coefficients were used to give different weights of CBR. We analyzed the performance of proposed weighted CBR based model compared to neural networks and pure CBR based model empirically and found that the proposed weighted CBR based model showed more superior performance than pure CBR model. Imbalanced data is a common problem to build data mining model to classify a class with real data such as bankruptcy prediction, intrusion detection, fraud detection, churn management, and response modeling. Imbalanced data means that the number of instance in one class is remarkably small or large compared to the number of instance in other classes. The classification model such as response modeling has a lot of trouble to recognize the pattern from data through learning because the model tends to ignore a small number of classes while classifying a large number of classes correctly. To resolve the problem caused from imbalanced data distribution, sampling method is one of the most representative approach. The sampling method could be categorized to under sampling and over sampling. However, CBR is not sensitive to data distribution because it doesn't learn from data unlike machine learning algorithm. In this study, we investigated the robustness of our proposed model while changing the ratio of response customers and nonresponse customers to the promotion program because the response customers for the suggested promotion is always a small part of nonresponse customers in the real world. We simulated the proposed model 100 times to validate the robustness with different ratio of response customers to response customers under the imbalanced data distribution. Finally, we found that our proposed CBR based model showed superior performance than compared models under the imbalanced data sets. Our study is expected to improve the performance of response model for the promotion program with CBR under imbalanced data distribution in the real world.

• Annual Conference on Human and Language Technology
• /
• 2018.10a
• /
• pp.361-363
• /
• 2018

FAQ 시스템은 주어진 질문과 가장 유사한 질의를 찾아 이에 대한 답을 제공하는 시스템이다. 질의 간의 유사도를 측정하기 위해 문장을 벡터로 표현하며 일반적으로 TFIDF, Okapi BM25와 같은 방법으로 계산한 단어 가중치 벡터를 이용하여 문장을 표현한다. 하지만 단어 가중치 벡터는 어휘적 정보를 표현하는데 유용한 반면 단어의 의미적인(semantic) 정보는 표현하기 어렵다. 본 논문에서는 이를 보완하고자 딥러닝을 이용한 문장 임베딩을 구축하고 단어 가중치 벡터와 문장 임베딩을 조합한 문장 유사도 계산 모델을 제안한다. 또한 문장 임베딩 구현 시 self-attention 기법을 적용하여 문장 내 중요한 부분에 가중치를 주었다. 실험 결과 제안하는 유사도 계산 모델은 비교 모델에 비해 모두 높은 성능을 보였고 self-attention을 적용한 실험에서는 추가적인 성능 향상이 있었다.