• Journal of Information Processing Systems
• /
• v.16 no.4
• /
• pp.795-808
• /
• 2020

Due to the poor evaluation indexes such as detection accuracy and recall rate when Yolov3 network detects aircraft in remote sensing images, in this paper, we propose a remote sensing image aircraft detection method based on machine vision. In order to improve the target detection effect, the Inception module was introduced into the Yolov3 network structure, and then the data set was cluster analyzed using the k-means algorithm. In order to obtain the best aircraft detection model, on the basis of our proposed method, we adjusted the network parameters in the pre-training model and improved the resolution of the input image. Finally, our method adopted multi-scale training model. In this paper, we used remote sensing aircraft dataset of RSOD-Dataset to do experiments, and finally proved that our method improved some evaluation indicators. The experiment of this paper proves that our method also has good detection and recognition ability in other ground objects.

• Electronics and Telecommunications Trends
• /
• v.34 no.1
• /
• pp.98-110
• /
• 2019

Machine learning has been applied to medical imaging and has shown an excellent recognition rate. Recently, there has been much interest in preventive medicine. If data are accessible, machine learning packages can be used easily in digital healthcare fields. However, it is necessary to prepare the data in advance, and model evaluation and tuning are required to construct a reliable model. On average, these processes take more than 80% of the total effort required. In this study, we describe the basic concepts of machine learning, pre-processing and visualization of datasets, feature engineering for reliable models, model evaluation and tuning, and the latest trends in popular machine learning frameworks. Finally, we survey a explainable machine learning analysis tool and will discuss the future direction of machine learning.

• Journal of Convergence for Information Technology
• /
• v.8 no.5
• /
• pp.101-106
• /
• 2018

Software are being studied to register and manage assets. And Methods for evaluating software systems have been based on subjective evaluation criteria. We propose an evaluation model for evaluating complex assets obtained from the complexity measurement of the preceding asset management system. We used scales to measure and provide logical complexity to measure the complexity of our architectural assets. And we used a method to evaluate whether it expresses attribute value of architecture asset. We have also built an evaluation model criterion for evaluating the usability of the asset data based on the ISO/IEC 25010 quality model characteristics of the SQuaRE Series. When the designers design the asset as a composite asset, the optional evaluation of the negative property that weights are assigned according to the characteristics of each asset is applied to secure the flexibility of the evaluation model.

• Journal of the Korea institute for structural maintenance and inspection
• /
• v.9 no.2
• /
• pp.163-172
• /
• 2005

This study aims to evaluate the feasibility of remodeling business by predicting the future price of apartment house after remodeling using Hedonic Price Model. The data concerning such 9 independent variables as location, unit size, unit plan, landscape, parking, the number of elapsed years after completion, number of units, mechanical performance, interior from 25 regions in Seoul metropolitan city were collected and evaluated by established evaluation criteria. The coefficients affecting the price of apartment unit were made by way of linear multi-regression and put into Hedonic Price Model. The feasibility evaluation model for apartment was made and verified by data of remodelled apartment. The predicted results using suggested evaluation model coincide with actual apartment market situations.

• Journal of the Korean Society of Clothing and Textiles
• /
• v.30 no.8
• /
• pp.1210-1221
• /
• 2006

The strength of the PB(Private Brand) is that it can eliminate intermediary in the distribution channel and thoroughly control the quality under its own name. This study has developed traditional studies on private brand proneness in approaching data processing and empirical point of view of a consumer's attitude buildup process on PB through 'recognition-attitude-action(behavioral attitude)'. The subjects of this study are consumers in their $20s{\sim}40s$ who are main customer groups of PBs. A screening process has taken place to select consumers with purchasing experiences of retailor PBs. The data is analyzed by 'Structural Equation Modeling' of Amos 5.0 to verify consumer attitude formation model on private apparel brand. The results generated from this study are as follows: First, the proposed consumer attitude model on private apparel brand consists of store evaluation, experiential product evaluation, cognitive product evaluation, hedonic attitude, utilitarian attitude and purchase intention. Second, not only positively influence on utilitarian attitude but hedonic attitude can arouse positive emotional reaction of a consumer. Third, the store evaluation is ahead of the product evaluation because PB is more related to the image of a store. The influence of the store on PB is relatively stronger when compared with NB.

• Journal of the Korean Operations Research and Management Science Society
• /
• v.26 no.1
• /
• pp.109-116
• /
• 2001

The detection of corporate failures is a subject that has been particularly amenable to cross-sectional financial ratio analysis. In most of firms, however, the financial data are available over past years. Because of this, a model utilizing these longitudinal data could provide useful information on the prediction of bankruptcy. To correctly reflect the longitudinal and firm-specific data, the generalized linear model with assuming the first order AR(autoregressive) process is proposed. The method is motivated by the clinical research that several characteristics are measured repeatedly from individual over the time. The model is compared with several other predictive models to evaluate the performance. By using the financial data from manufacturing corporations in the Korea Stock Exchange (KSE) list, we will discuss some experiences learned from the procedure of sampling scheme, variable transformation, imputation, variable selection, and model evaluation. Finally, implications of the model with repeated measurement and future direction of research will be discussed.

• Korean Journal of Legislative Studies
• /
• v.23 no.2
• /
• pp.147-177
• /
• 2017

Although a number of empirical studies found that political ideology plays a significant role in Korean elections, they entirely rely on cross-sectional data analysis. In contrast to previous research, this study investigates the effects of ideology in the 2012 Korean presidential election through standard panel data analysis. Specifically, using "EAI Panel Study, 2012", the effects of ideology on both candidate evaluation and vote choice were examined via fixed effects, random effects, and pooled regression analysis. And the results from applying the two most popular models of ideological voting, the proximity model and the directional change model were also compared. The results show that candidate evaluations and vote choice during the election (April, 2012- December, 2012) were significantly influenced by the ideological difference between voters and candidates, independent from partisanship and other standard socio-demographic factors. And this ideological voting during the election seems better captured by the directional change model than by the proximity model.

• Journal of Korea Multimedia Society
• /
• v.21 no.7
• /
• pp.779-786
• /
• 2018

In this paper, we are applied a nerve network to allow for the reflection of data learning methods in their overall forms by using cluster data rather than data learning by the stages and then selected a nerve network model and analyzed its variables through learning by the cluster. The CkLR algorithm was proposed to analyze the reaction variables of clustering outcomes through an approach to the initialization of K-means clustering and build a model to assess the prediction rate of clustering and the accuracy rate of prediction in case of new data inputs. The performance evaluation results show that the accuracy rate of test data by the class was over 92%, which was the mean accuracy rate of the entire test data, thus confirming the advantages of a specialized structure found in the proposed learning nerve network by the class.

• KIPS Transactions on Software and Data Engineering
• /
• v.10 no.9
• /
• pp.375-384
• /
• 2021

Deep learning techniques have been proven to have high performance in image processing and are applied in various fields. The most widely used methods for validating a deep learning model include a holdout verification method, a k-fold cross verification method, and a bootstrap method. These legacy methods consider the balance of the ratio between classes in the process of dividing the data set, but do not consider the ratio of various features that exist within the same class. If these features are not considered, verification results may be biased toward some features. Therefore, we propose a deep learning model validation method based on data feature coverage for image classification by improving the legacy methods. The proposed technique proposes a data feature coverage that can be measured numerically how much the training data set for training and validation of the deep learning model and the evaluation data set reflects the features of the entire data set. In this method, the data set can be divided by ensuring coverage to include all features of the entire data set, and the evaluation result of the model can be analyzed in units of feature clusters. As a result, by providing feature cluster information for the evaluation result of the trained model, feature information of data that affects the trained model can be provided.

• Smart Structures and Systems
• /
• v.11 no.6
• /
• pp.569-587
• /
• 2013

This paper evaluates the closed loop performance of the reaching law based discrete sliding mode controller with multisensor data fusion (MSDF) in real time, by controlling the first two vibrating modes of a piezo actuated structure. The vibration is measured using two homogeneous piezo sensors. The states estimated from sensors output are fused. Four fusion algorithms are considered, whose output is used to control the structural vibration. The controller is designed using a model identified through linear Recursive Least Square (RLS) method, based on ARX model. Improved vibration suppression is achieved with fused data as compared to single sensor. The experimental evaluation of the closed loop performance of sliding mode controller with data fusion applied to piezo actuated structure is the contribution in this work.