• 지능정보연구
• /
• 제29권4호
• /
• pp.325-345
• /
• 2023

본 연구는 온라인 리뷰를 이용하여 고객 만족도를 예측하는 새로운 접근 방식을 제안한다. LDA 주제 모델링과 결합된 RFE-SHAP 기능 선택 방법을 활용하여 고객 만족도에 큰 영향을 미치는 주요 기능을 식별하여 예측 분석을 개선했다. 먼저 Random Forest 알고리즘의 경우, 초기 28개 입력변수에서 14개의 변수를 최적 하위 집합으로 추출했다. 제안된 방법에서 Random Forest 모델의 성과는 84%로 확인 되었으며 변수가 많은 모델에서 흔히 발생하는 과적합을 방지하였다. 또한 품질, 착용감, 내구성 등과 같은 리뷰의 특정 요소들이 패션 산업 내에서 소비자 만족도를 증진시키는 중요한 역할을 한다는 사실을 밝혀냈다. 본 연구는 예측 결과를 설명할 때 선택한 각 기능이 고객 만족도에 어떻게 영향을 미치는지에 대한 자세한 설명을 제공하고 고객이 가장 중요하게 생각하는 측면에 대한 세부적인 보기를 제공한다. 본 연구의 공헌도는 다음과 같다. 첫째, 전자상거래 분석 분야 내에서 예측 모델링을 강화하고 특성 중심적인 접근법을 소개함으로써 방법론을 개선하였다. 이는 고객 만족도 예측의 정확도를 높일 뿐만 아니라 예측 모델에서의 변수 선택에 대한 새로운 접근을 제시한다. 둘째, 특히 의류 부문에서 전자상거래 플랫폼에 구체적인 통찰력을 제공한다. 품질, 사이즈, 내구성 등 고객 리뷰의 어떤 부분이 만족도에 가장 큰 영향을 미치는지 강조함으로써, 기업들이 제품과 서비스를 맞춤화 할 수 있는 전략적 방향을 제시한다. 이러한 목표 지향적인 개선은 고객의 쇼핑 경험을 개선하고, 만족도를 향상시키면서 충성도를 이끌어낼 수 있을 것으로 기대한다.

• ETRI Journal
• /
• 제32권2호
• /
• pp.292-301
• /
• 2010

Increasing convergence among heterogeneous radio networks is expected to be a key feature of future ubiquitous services. The convergence of radio networks in combination with dynamic spectrum allocation (DSA) could be a beneficial means to solve the growing demand for radio spectrum. DSA might enhance the spectrum utilization of involved radio networks to comply with user requirements for high-quality multimedia services. This paper proposes a simple spectrum allocation algorithm and presents an analytical model of dynamic spectrum resource allocation between two networks using a 4-D Markov chain. We argue that there may exist a break-even point for choosing whether or not to adopt DSA in a system. We point out certain circumstances where DSA is not a viable alternative. We also discuss the performance of DSA against the degree of resource sharing using the proposed analytical model and simulations. The presented analytical model is not restricted to DSA, and can be applied to a general resource sharing study.

• 제어로봇시스템학회논문지
• /
• 제18권12호
• /
• pp.1090-1095
• /
• 2012

In this paper, we propose a model-based fault diagnosis method applied to vibration data. The fault detection is performed by comparing estimated parameters with normal parameters and deciding if the observed changes can be explained satisfactorily in terms of noise or undermodelling. The key feature of this method is that it accounts for the effects of noise and model mismatch. And we aslo design a classifier for the fault isolation by applying the multiclass SVM (Support Vector Machine) to the estimated parameters. The proposed fault detection and isolation methods are applied to an engine vibration data to show a good performance. The proposed fault detection method is compared with a signal-based fault detection method through a performance analysis.

• 한국멀티미디어학회논문지
• /
• 제20권2호
• /
• pp.144-152
• /
• 2017

Among the technologies for implementing autonomous vehicles, advanced driver assistance system is a key technology to support driver's safe driving. In the technology using the vision sensor having a high utility, various preprocessing methods are used prior to feature extraction for lane detection. However, in the existing methods, the unnecessary lane candidates such as cars, lawns, and road separator in the road area are false positive. In addition, there are cases where the lane candidate itself can not be extracted in the area under the overpass, the lane within the dark shadow, the center lane of yellow, and weak lane. In this paper, we propose an efficient preprocessing method using k-means clustering for image division and the HSV color model. When the proposed preprocessing method is applied, the true positive region is maximally maintained during the lane detection and many false positive regions are removed.

• 산업융합연구
• /
• 제20권5호
• /
• pp.61-68
• /
• 2022

최근에는 발전하는 정보통신 기술을 이용하여 악의적인 코드들이 제작되고 있고 이를 기존 탐지 시스템으로는 탐지하는게 역부족인 실정이다. 이러한 지능적이고 악의적인 코드를 정확하고 효율성 있게 탐지하고 대응하기 위해서는 지능적 탐지 모델이 필요하다. 그리고, 탐지 성능을 최대로 높이기 위해서는 악의적인 코드의 주요 특징 정보 집합으로 훈련하는 것이 중요하다. 본 논문에서는 지능적 탐지 모델을 설계하고 모델 훈련에 필요한 데이터를 변환, 차원축소, 특징 선택 단계를 거쳐 주요 특징 정보 집합으로 생성하는 기법을 제안하였다. 그리고 이를 기반으로 악의적인 코드별로 주요 특징 정보를 분류하였다. 또한, 분류된 특징 정보들을 기반으로 변형되거나 새로 등장하는 악의적인 코드를 분석하고 탐지하는데 사용할 수 있는 공통 특징 정보를 도출하였다. 제안된 탐지 모델은 제한된 수의 특성 정보로 학습하여 악의적인 코드를 탐지하기에 탐지 시간과 대응이 빨리 이루어져 피해를 크게 줄일 수 있다. 그리고, 성능 평가 결과값은 학습 알고리즘에 따라 약간 차이가 나지만 악의적인 코드 대부분을 탐지할 수 있음을 평가로 알 수 있었다.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제15권12호
• /
• pp.4326-4344
• /
• 2021

The SinGAN is one of generative adversarial networks that can be trained on a single nature image. It has poor ability to learn more global features from nature image, and losses much local detail information when it generates arbitrary size image sample. To solve the problem, a non-linear function is firstly proposed to control downsampling ratio that is ratio between the size of current image and the size of next downsampled image, to increase the ratio with increase of the number of downsampling. This makes the low-resolution images obtained by downsampling have higher proportion in all downsampled images. The low-resolution images usually contain much global information. Therefore, it can help the model to learn more global feature information from downsampled images. Secondly, the attention mechanism is introduced to the generative network to increase the weight of effective image information. This can make the network learn more local details. Besides, in order to make the output image more natural, the TVLoss function is introduced to the loss function of SinGAN, to reduce the difference between adjacent pixels and smear phenomenon for the output image. A large number of experimental results show that our proposed model has better performance than other methods in generating random samples with fixed size and arbitrary size, image harmonization and editing.

• Structural Monitoring and Maintenance
• /
• 제5권1호
• /
• pp.67-78
• /
• 2018

The stability of cutting slope influences the safety of railway operation, and how to identify the stability of the slope quickly and determine the rational monitoring plan is a pressing problem at present. In this study, the attribute recognition model of risk assessment for high cutting slope stability in the heavy haul railway is established based on attribute mathematics theory, followed by the consequent monitoring scheme design. Firstly, based on comprehensive analysis on the risk factors of heavy haul railway loess slope, collapsibility, tectonic feature, slope shape, rainfall, vegetation conditions, train speed are selected as the indexes of the risk assessment, and the grading criteria of each index is established. Meanwhile, the weights of the assessment indexes are determined by AHP judgment matrix. Secondly, The attribute measurement functions are given to compute attribute measurement of single index and synthetic attribute, and the attribute recognition model was used to assess the risk of a typical heavy haul railway loess slope, Finally, according to the risk assessment results, the monitoring content and method of this loess slope were determined to avoid geological disasters and ensure the security of the railway infrastructure. This attribute identification- risk assessment- monitoring design mode could provide an effective way for the risk assessment and control of heavy haul railway in the loess plateau.

• International Journal of Aeronautical and Space Sciences
• /
• 제14권3호
• /
• pp.247-255
• /
• 2013

The flight control of re-entry vehicles poses a challenge to conventional gain-scheduled flight controllers due to the widely spread aerodynamic coefficients. In addition, a wide range of uncertainties in disturbances must be accommodated by the control system. This paper presents the design of a roll channel controller for a non-axisymmetric reentry vehicle model using the trajectory linearization control (TLC) method. The dynamic equations of a moving mass system and roll control model are established using the Lagrange method. Nonlinear tracking and decoupling control by trajectory linearization can be viewed as the ideal gain-scheduling controller designed at every point along the flight trajectory. It provides robust stability and performance at all stages of the flight without adjusting controller gains. It is this "plug-and-play" feature that is highly preferred for developing, testing and routine operating of the re-entry vehicles. Although the controller is designed only for nominal aerodynamic coefficients, excellent performance is verified by simulation for wind disturbances and variations from -30% to +30% of the aerodynamic coefficients.

• 한국측량학회지
• /
• 제16권1호
• /
• pp.139-147
• /
• 1998

국토에 대한 공간정보의 기본도로서 다양한 사용자의 요구사항을 충족시킬 수 있는 수치지도를 효율적으로 제작 및 유지관리하기 위해서는 수치지도의 데이터 모델과 구조 등 수치지도 데이터 구축에 대한 연구가 중점적으로 이루어져야 한다. 본 연구에서는 단순 지도제작을 목적으로 구축된 기존 수치지도의 데이터 모델에 대한 문제점을 파악하여, 개선된 수치지도 데이터 모델의 개념적ㆍ논리적 모델을 제시하였다. 제시된 수치지도 데이터 모델은 기존에 제작된 수치지도 데이터의 사용을 전제로 설계되었으며, 다양한 사용자의 요구사항을 반영하기 위해서 데이터 모델의 유연성에 중점을 두었고, 필요에 따라서 수정이 용이하도록 단계별 접근방법에 의해 설계되었다. 또한 제시된 수치지도 데이터 모델은 향후 객체 지향형 모델로 나아가기 위한 중간 단계의 데이터 모델이다.

• The Korean Journal of Pain
• /
• 제32권2호
• /
• pp.120-128
• /
• 2019

Background: We aimed to explore the American College of Rheumatology (ACR) 1990 and 2011 fibromyalgia (FM) classification criteria's items and the components of Fibromyalgia Impact Questionnaire (FIQ) to identify features best discriminating FM features. Finally, we developed a combined FM diagnostic (C-FM) model using the FM's key features. Methods: The means and frequency on tender points (TPs), ACR 2011 components and FIQ items were calculated in the FM and non-FM (osteoarthritis [OA] and non-OA) patients. Then, two-step multiple logistic regression analysis was performed to order these variables according to their maximal statistical contribution in predicting group membership. Partial correlations assessed their unique contribution, and two-group discriminant analysis provided a classification table. Using receiver operator characteristic analyses, we determined the sensitivity and specificity of the final model. Results: A total of 172 patients with FM, 75 with OA and 21 with periarthritis or regional pain syndromes were enrolled. Two steps multiple logistic regression analysis identified 8 key features of FM which accounted for 64.8% of variance associated with FM group membership: lateral epicondyle TP with variance percentages (36.9%), neck pain (14.5%), fatigue (4.7%), insomnia (3%), upper back pain (2.2%), shoulder pain (1.5%), gluteal TP (1.2%), and FIQ fatigue (0.9%). The C-FM model demonstrated a 91.4% correct classification rate, 91.9% for sensitivity and 91.7% for specificity. Conclusions: The C-FM model can accurately detect FM patients among other pain disorders. Re-inclusion of TPs along with saving of FM main symptoms in the C-FM model is a unique feature of this model.