• The Journal of the Korea Contents Association
• /
• v.21 no.8
• /
• pp.1-9
• /
• 2021

Deep learning in computer vision has made accelerated improvement over a short period but large-scale learning data and computing power are still essential that required time-consuming trial and error tasks are involved to derive an optimal network model. In this study, we propose a similar image classification performance improvement method based on CR (Confusion Rate) that considers only the characteristics of the data itself regardless of network optimization or data reinforcement. The proposed method is a technique that improves the performance of the deep learning model by calculating the CRs for images in a dataset with similar characteristics and reflecting it in the weight of the Loss Function. Also, the CR-based recognition method is advantageous for image identification with high similarity because it enables image recognition in consideration of similarity between classes. As a result of applying the proposed method to the Resnet18 model, it showed a performance improvement of 0.22% in HanDB and 3.38% in Animal-10N. The proposed method is expected to be the basis for artificial intelligence research using noisy labeled data accompanying large-scale learning data.

• Journal of the Korea Society of Computer and Information
• /
• v.27 no.7
• /
• pp.93-100
• /
• 2022

In the era of information overload on the Internet, the recommendation system, which is an indispensable function, is a service that recommends products that a user may prefer, and has been successfully provided in various commercial sites. Recently, studies to reflect the rating time of items to improve the performance of collaborative filtering, a representative recommendation technique, are active. The core idea of these studies is to generate the recommendation list by giving an exponentially lower weight to the items rated in the past. However, this has a disadvantage in that a time function is uniformly applied to all items without considering changes in users' preferences according to the characteristics of the items. In this study, we propose a time-aware collaborative filtering technique from a completely different point of view by developing a new similarity measure that integrates the change in similarity values between items over time into a weighted sum. As a result of the experiment, the prediction performance and recommendation performance of the proposed method were significantly superior to the existing representative time aware methods and traditional methods.

• Journal of Korean Tunnelling and Underground Space Association
• /
• v.25 no.3
• /
• pp.187-200
• /
• 2023

This study analyzed the seismic fragility of bored tunnels based on their surrounding conditions and suggested a representative seismic fragility model. By analyzing the existed seismic fragility models developed for bored tunnels, we developed weighted combination models for each surrounding conditions, such as ground conditions and depth of the tunnel. The seismic fragility curves use the peak ground acceleration (PGA) as a parameter. When the PGA was 0.3 g, the probability of damage exceeding minor or slight damage was 20% for depth of 50 m or less, 10% for depth between 50 m and 100 m, and 3% for depth of 100 m or more. It was also found that the probability of damage was higher for the same PGA and depth when the surrounding ground was rock rather than soil. The probability of damage decreases as the depth increase. This study is expected to be used for developing a comprehensive seismic fragility function for tunnels in the future.

• Journal of the Institute of Convergence Signal Processing
• /
• v.23 no.4
• /
• pp.185-193
• /
• 2022

This paper proposes an adaptive sliding mode control with radial basis function neural network(RBFNN) scheme to enhance the performance of position and attitude tracking control of quadrotor UAV. The RBFNN is utilized on the approximation of nonlinear function in the UAV dynmic model and the weights of the RBFNN are adjusted online according to adaptive law from the Lyapunov stability analysis to ensure the state hitting the sliding surface and sliding along it. In order to compensate the network approximation error and eliminate the existing chattering problems, the sliding mode control term is adjusted by adaptive laws, which can enhance the robust performance of the system. The simulation results of the proposed control method confirm the effectiveness of the proposed controller which applied for a nonlinear quadrotor UAV is presented. Form the results, it's shown that the developed control system is achieved satisfactory control performance and robustness.

• The Journal of the Acoustical Society of Korea
• /
• v.41 no.6
• /
• pp.647-654
• /
• 2022

Although the fields in which is utilized children's speech recognition is on the rise, the lack of quality data is an obstacle to improving children's speech recognition performance. This paper proposes a new method for improving children's speech recognition performance by additionally using adult speech data. The proposed method is a transformer based domain adversarial training using dynamically weighted loss to effectively address the data imbalance gap between age that grows as the amount of adult training data increases. Specifically, the degree of class imbalance in the mini-batch during training was quantified, and the loss function was defined and used so that the smaller the data, the greater the weight. Experiments validate the utility of proposed domain adversarial training following asymmetry between adults and children training data. Experiments show that the proposed method has higher children's speech recognition performance than traditional domain adversarial training method under all conditions in which asymmetry between age occurs in the training data.

• Proceedings of the Korea Water Resources Association Conference
• /
• 2021.06a
• /
• pp.429-429
• /
• 2021

2019년 발생한 인천광역시 붉은 수돗물 사태로 급수구역에 포함된 26만 1천 세대, 63만 5천 명이 직·간접적인 피해를 입은 바 있다. 경제적 피해액으로 추정할 경우 최소 1,280억 원 이상으로 보고된 바 있으며, 이와 같은 상수관망의 수질사고 확산은 장기간 동안 시민의 건강과 생활환경 수준을 저하시킨다. 따라서 상수도시스템의 수질사고확산 모델링 및 방지기술을 통한 수질안전성의 재확인이 필요하며, 이것은 상수도시스템의 지속가능성을 높여 국민이 체감하는 물 환경 수준 제고에 기여가 가능하다. 관망 내 수질해석을 직접적으로 수행하는 모델은 국외적으로 다양하게 개발(PODDS, EPANET-MSX, EPANET2.2 등)된 바 있으나 검·보정을 위한 수질측정 자료 부족 등으로 적용이 제한적이라는 한계가 현재에도 존재한다. 이를 보완하기 위해 수질자료에 비해 그 양이 많고 획득방법이 상대적으로 수월한 수리학적 계측자료 및 해석결과를 활용한 관로 내 체류시간 등을 활용한 연구가 수행된 바 있다. 그러나 이와 같은 수리학적 해석 결과를 활용하는 경우에도 계측자료를 기반으로 한 수리학적 검·보정은 필수적이라 할 수 있다. 본 연구에서는 관로 내 체류시간에 직접적인 영향을 미치는 유량 및 유속자료를 중심으로 수리학적 관망해석의 결과를 최적 검·보정하는 방법론을 제안하였다. 기존 상수관망 수리해석의 검·보정은 일부 지점에서 수압을 측정하고, 수리해석 결과로 도출되는 해당 지점의 수압이 실측된 결과와 유사하도록 관로의 유속계수를 적절히 보정하는 형태로 진행되었다. 그러나 본 연구에서는 관로유량과 유속자료의 목적함수 내 가중치를 수압자료보다 크게 설정하여 체류시간 중심의 검·보정이 수행될 수 있도록 하였으며, 검·보정 대상인자 역시 대수용가의 수요량, 수요패턴, 그리고 관로유속계수로 확장된 모형을 구축하였다. 최적화 기법으로는 메타휴리스틱 기법중 하나인 화음탐색법을 활용하였다. EPANET 2.2 Toolkit과 Visual Basic .Net을 연계하여 프로그래밍하였으며, 개발된 모형을 실제 지방상수도 시스템에 적용하여 분석하였다.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2022.05a
• /
• pp.520-523
• /
• 2022

일본 우주항공연구개발기구(Japan Aerospace Exploration Agency, JAXA)는 2007년부터 2017년까지 달 탐사선 셀레네(Selenological and Engineering Explorer, SelEnE)가 관측한 데이터를 수집하고, 연구했다. JAXA는 지구 상층 대기에 존재하는 산소가 자기장의 꼬리 부분에 실려 달로 이동한다는 사실을 발견했다. 하지만 이 연구는 아직 진행 중이며 달의 산화 과정 규명에 추가 연구가 필요하다. 본 논문에서는 생성적 적대 신경망(Generative Adversarial Networks, GAN)으로 달 분화구의 영구 그림자 영역을 제거하고, 물과 얼음을 발견하여 선행 연구의 완성도를 향상하고자 한다. 실험에 사용할 모델은 CIPS(Conditionally Independent Pixel Synthesis)다. CIPS는 실제 같은 영상을 고해상도로 합성한다. 합성할 데이터의 최적인 가중치 초기화 및 파라미터 갱신 방법, 활성 함수 조합은 실험을 통해 확인한다. 필요에 따라 앙상블 학습을 할 수도 있다. 성능평가는 FID(Frechet Inception Distance), 정밀도, 재현율을 사용한다. 제안한 방법은 진행 중인 연구의 시간과 비용을 절약하고, 인과관계를 더욱 명확히 밝히는 데 도움 될 수 있다고 사료된다.

• The Journal of the Acoustical Society of Korea
• /
• v.28 no.3
• /
• pp.268-272
• /
• 2009

In this paper, we propose a Gaussian Mixture Model (GMM) based incremental robust adaptation with a forgetting factor for the speaker verification. Speaker recognition system uses a speaker model adaptation method with small amounts of data in order to obtain a good performance. However, a conventional adaptation method has vulnerable to the outlier from the irregular utterance variations and the presence noise, which results in inaccurate speaker model. As time goes by, a rate in which new data are adapted to a model is reduced. The proposed algorithm uses an incremental robust adaptation in order to reduce effect of outlier and use forgetting factor in order to maintain adaptive rate of new data on GMM based speaker model. The incremental robust adaptation uses a method which registers small amount of data in a speaker recognition model and adapts a model to new data to be tested. Experimental results from the data set gathered over seven months show that the proposed algorithm is robust against outliers and maintains adaptive rate of new data.

• Journal of the Institute of Electronics and Information Engineers
• /
• v.53 no.6
• /
• pp.110-121
• /
• 2016

The primary aim of the color constancy algorithm is to estimate illuminant chromaticity. There are various statistical-based, learning-based and combinational-based color constancy algorithms already exist. However, the statistical-based algorithms can only perform well on images that satisfy certain assumptions, learning-based methods are complex methods that require proper preprocessing and training data, and combinational-based methods depend on either pre-determined or dynamically varying weights, which are difficult to determine and prone to error. Therefore, this paper presents a new optimization based illuminant estimation method which is free from complex preprocessing and can estimate the illuminant under different environmental conditions. A strong color cast always has an odd standard deviation value in one of the RGB channels. Based on this observation, a cost function called the degree of illuminant tinge(DIT) is proposed to determine the quality of illuminant color-calibrated images. This DIT is formulated in such a way that the image scene under standard illuminant (d65) has lower DIT value compared to the same scene under different illuminant. Here, a swarm intelligence based particle swarm optimizer(PSO) is used to find the optimum illuminant of the given image that minimizes the degree of illuminant tinge. The proposed method is evaluated using real-world datasets and the experimental results validate the effectiveness of the proposed method.

• Journal of the Korea Concrete Institute
• /
• v.26 no.1
• /
• pp.13-21
• /
• 2014

To quantitatively evaluate the influence of ground granulated blast-furnace slag (GGBS) as a supplementary cementitious material on the life-cycle environmental impact of concrete, a comprehensive database including 3395 laboratory mixes and 1263 plant mixes was analyzed. The life-cycle assesment studied for the environmental impact of concrete can be summarized as follows: 1) the system boundary considered was from cradle to pre-construction; 2) Korea life-cycle inventories were primarily used to assess the environmental loads in each phase of materials, transportation and production of concrete; and 3) the environmental loads were quantitatively converted into environmental impact indicators through categorization, characterization, normalization and weighting process. The life-cycle environmental impacts of concrete could be classified into three categories including global warming, photochemical oxidant creation and abiotic resource depletion. Furthermore, these environmental impacts of concrete was significantly governed by the unit content of ordinary portland cement (OPC) and decreased with the increase of the replacement level of GGBS. As a result, simple equations to assess the environmental impact indicators could be formulated as a function of the unit content of binder and replacement level of GGBS.