• Journal of Intelligence and Information Systems
• /
• v.24 no.1
• /
• pp.205-225
• /
• 2018

Convolutional Neural Network (ConvNet) is one class of the powerful Deep Neural Network that can analyze and learn hierarchies of visual features. Originally, first neural network (Neocognitron) was introduced in the 80s. At that time, the neural network was not broadly used in both industry and academic field by cause of large-scale dataset shortage and low computational power. However, after a few decades later in 2012, Krizhevsky made a breakthrough on ILSVRC-12 visual recognition competition using Convolutional Neural Network. That breakthrough revived people interest in the neural network. The success of Convolutional Neural Network is achieved with two main factors. First of them is the emergence of advanced hardware (GPUs) for sufficient parallel computation. Second is the availability of large-scale datasets such as ImageNet (ILSVRC) dataset for training. Unfortunately, many new domains are bottlenecked by these factors. For most domains, it is difficult and requires lots of effort to gather large-scale dataset to train a ConvNet. Moreover, even if we have a large-scale dataset, training ConvNet from scratch is required expensive resource and time-consuming. These two obstacles can be solved by using transfer learning. Transfer learning is a method for transferring the knowledge from a source domain to new domain. There are two major Transfer learning cases. First one is ConvNet as fixed feature extractor, and the second one is Fine-tune the ConvNet on a new dataset. In the first case, using pre-trained ConvNet (such as on ImageNet) to compute feed-forward activations of the image into the ConvNet and extract activation features from specific layers. In the second case, replacing and retraining the ConvNet classifier on the new dataset, then fine-tune the weights of the pre-trained network with the backpropagation. In this paper, we focus on using multiple ConvNet layers as a fixed feature extractor only. However, applying features with high dimensional complexity that is directly extracted from multiple ConvNet layers is still a challenging problem. We observe that features extracted from multiple ConvNet layers address the different characteristics of the image which means better representation could be obtained by finding the optimal combination of multiple ConvNet layers. Based on that observation, we propose to employ multiple ConvNet layer representations for transfer learning instead of a single ConvNet layer representation. Overall, our primary pipeline has three steps. Firstly, images from target task are given as input to ConvNet, then that image will be feed-forwarded into pre-trained AlexNet, and the activation features from three fully connected convolutional layers are extracted. Secondly, activation features of three ConvNet layers are concatenated to obtain multiple ConvNet layers representation because it will gain more information about an image. When three fully connected layer features concatenated, the occurring image representation would have 9192 (4096+4096+1000) dimension features. However, features extracted from multiple ConvNet layers are redundant and noisy since they are extracted from the same ConvNet. Thus, a third step, we will use Principal Component Analysis (PCA) to select salient features before the training phase. When salient features are obtained, the classifier can classify image more accurately, and the performance of transfer learning can be improved. To evaluate proposed method, experiments are conducted in three standard datasets (Caltech-256, VOC07, and SUN397) to compare multiple ConvNet layer representations against single ConvNet layer representation by using PCA for feature selection and dimension reduction. Our experiments demonstrated the importance of feature selection for multiple ConvNet layer representation. Moreover, our proposed approach achieved 75.6% accuracy compared to 73.9% accuracy achieved by FC7 layer on the Caltech-256 dataset, 73.1% accuracy compared to 69.2% accuracy achieved by FC8 layer on the VOC07 dataset, 52.2% accuracy compared to 48.7% accuracy achieved by FC7 layer on the SUN397 dataset. We also showed that our proposed approach achieved superior performance, 2.8%, 2.1% and 3.1% accuracy improvement on Caltech-256, VOC07, and SUN397 dataset respectively compare to existing work.

• Journal of Intelligence and Information Systems
• /
• v.18 no.4
• /
• pp.1-17
• /
• 2012

Personalized smart devices such as smartphones and smart pads are widely used. Unlike traditional feature phones, theses smart devices allow users to choose a variety of functions, which support not only daily experiences but also business operations. Actually, there exist a huge number of applications accessible by smart device users in online and mobile application markets. Users can choose apps that fit their own tastes and needs, which is impossible for conventional phone users. With the increase in app demand, the tastes and needs of app users are becoming more diverse. To meet these requirements, numerous apps with diverse functions are being released on the market, which leads to fierce competition. Unlike offline markets, online markets have a limitation in that purchasing decisions should be made without experiencing the items. Therefore, online customers rely more on item-related information that can be seen on the item page in which online markets commonly provide details about each item. Customers can feel confident about the quality of an item through the online information and decide whether to purchase it. The same is true of online app markets. To win the sales competition against other apps that perform similar functions, app developers need to focus on writing app descriptions to attract the attention of customers. If we can measure the effect of app descriptions on sales without regard to the app's price and quality, app descriptions that facilitate the sale of apps can be identified. This study intends to provide such a quantitative result for app developers who want to promote the sales of their apps. For this purpose, we collected app details including the descriptions written in Korean from one of the largest app markets in Korea, and then extracted keywords from the descriptions. Next, the impact of the keywords on sales performance was measured through our econometric model. Through this analysis, we were able to analyze the impact of each keyword itself, apart from that of the design or quality. The keywords, comprised of the attribute and evaluation of each app, are extracted by a morpheme analyzer. Our model with the keywords as its input variables was established to analyze their impact on sales performance. A regression analysis was conducted for each category in which apps are included. This analysis was required because we found the keywords, which are emphasized in app descriptions, different category-by-category. The analysis conducted not only for free apps but also for paid apps showed which keywords have more impact on sales performance for each type of app. In the analysis of paid apps in the education category, keywords such as 'search+easy' and 'words+abundant' showed higher effectiveness. In the same category, free apps whose keywords emphasize the quality of apps showed higher sales performance. One interesting fact is that keywords describing not only the app but also the need for the app have asignificant impact. Language learning apps, regardless of whether they are sold free or paid, showed higher sales performance by including the keywords 'foreign language study+important'. This result shows that motivation for the purchase affected sales. While item reviews are widely researched in online markets, item descriptions are not very actively studied. In the case of the mobile app markets, newly introduced apps may not have many item reviews because of the low quantity sold. In such cases, item descriptions can be regarded more important when customers make a decision about purchasing items. This study is the first trial to quantitatively analyze the relationship between an item description and its impact on sales performance. The results show that our research framework successfully provides a list of the most effective sales key terms with the estimates of their effectiveness. Although this study is performed for a specified type of item (i.e., mobile apps), our model can be applied to almost all of the items traded in online markets.

• Journal of Intelligence and Information Systems
• /
• v.26 no.2
• /
• pp.1-25
• /
• 2020

In this paper, we suggest an application system architecture which provides accurate, fast and efficient automatic gasometer reading function. The system captures gasometer image using mobile device camera, transmits the image to a cloud server on top of private LTE network, and analyzes the image to extract character information of device ID and gas usage amount by selective optical character recognition based on deep learning technology. In general, there are many types of character in an image and optical character recognition technology extracts all character information in an image. But some applications need to ignore non-of-interest types of character and only have to focus on some specific types of characters. For an example of the application, automatic gasometer reading system only need to extract device ID and gas usage amount character information from gasometer images to send bill to users. Non-of-interest character strings, such as device type, manufacturer, manufacturing date, specification and etc., are not valuable information to the application. Thus, the application have to analyze point of interest region and specific types of characters to extract valuable information only. We adopted CNN (Convolutional Neural Network) based object detection and CRNN (Convolutional Recurrent Neural Network) technology for selective optical character recognition which only analyze point of interest region for selective character information extraction. We build up 3 neural networks for the application system. The first is a convolutional neural network which detects point of interest region of gas usage amount and device ID information character strings, the second is another convolutional neural network which transforms spatial information of point of interest region to spatial sequential feature vectors, and the third is bi-directional long short term memory network which converts spatial sequential information to character strings using time-series analysis mapping from feature vectors to character strings. In this research, point of interest character strings are device ID and gas usage amount. Device ID consists of 12 arabic character strings and gas usage amount consists of 4 ~ 5 arabic character strings. All system components are implemented in Amazon Web Service Cloud with Intel Zeon E5-2686 v4 CPU and NVidia TESLA V100 GPU. The system architecture adopts master-lave processing structure for efficient and fast parallel processing coping with about 700,000 requests per day. Mobile device captures gasometer image and transmits to master process in AWS cloud. Master process runs on Intel Zeon CPU and pushes reading request from mobile device to an input queue with FIFO (First In First Out) structure. Slave process consists of 3 types of deep neural networks which conduct character recognition process and runs on NVidia GPU module. Slave process is always polling the input queue to get recognition request. If there are some requests from master process in the input queue, slave process converts the image in the input queue to device ID character string, gas usage amount character string and position information of the strings, returns the information to output queue, and switch to idle mode to poll the input queue. Master process gets final information form the output queue and delivers the information to the mobile device. We used total 27,120 gasometer images for training, validation and testing of 3 types of deep neural network. 22,985 images were used for training and validation, 4,135 images were used for testing. We randomly splitted 22,985 images with 8:2 ratio for training and validation respectively for each training epoch. 4,135 test image were categorized into 5 types (Normal, noise, reflex, scale and slant). Normal data is clean image data, noise means image with noise signal, relfex means image with light reflection in gasometer region, scale means images with small object size due to long-distance capturing and slant means images which is not horizontally flat. Final character string recognition accuracies for device ID and gas usage amount of normal data are 0.960 and 0.864 respectively.

• Korea Journal of Hospital Management
• /
• v.8 no.1
• /
• pp.112-134
• /
• 2003

The purpose of this study is to analyze the consumer's expectation before the health care service and the consumer's satisfaction after it. The participants of the study are inpatients in a general hospital located in Seoul. The resources were collected from the self-administration questionnaire survey run parallel with face to face interview. In order to measure the degree of the consumer's expectation, 349 samples were collected from the first questionnaire survey on the date of admission to the hospital. The second questionnaire survey was carried out on the date of discharge to the hospital with the participants responding to the first questionnaire survey. There are 154 samples collected from this survey. The results from the analysis of these resources are as follow. First, the survey shows that one of the highest consumers' expectations was about the generosity, kindliness and sincerity from the staff at the hospital, specially from doctors. Second, according to the analysis of the factors affecting the expectations of the consumers, with regard to path of admission to a hospital relating to patient's features, outpatient who gets into a hospital expected good medical care much more than the other patients. In regard of doctor's features, patients usually and highly expect good medical care from doctors who have good carrier and much experience. Third, according to the second questionnaire survey, what patients are satisfied most with is about the generosity and sincerity from staff at a hospital, especially from doctors and their gem attitudes. The results from survey show that the differences among the degree of consumers' satisfaction are very variable, depending on surrounding environments and facilities. The only fact that expectation didn't meet with satisfaction appeared to the case about technology and skill of medical care and the case about updated medical skills and equipments. Fourth, comparing the degree of expectation with the degree of satisfaction of consumers, correlative analysis was concerned significantly and specifically about the part of overall cleanliness relating to facilities and surrounding environments, the items about medical examination and test plan procedure relating to skill of medical care, professional specialties and convenience for procedure, and the items about satisfying explanations and concern about patients from doctors relating to staff's generosity and sincerity. Fifth, the analysis of the factors affecting the degree of how much patients are satisfied with shows that relating to sociodemographical features, patients are not satisfied with the case when the time and process of medical treatment are getting longer. It is surveyed that consumer were satisfied with the motivation to visit a hospital and the insurance type in patient's feature and so were the medical department and the factor of the degree of the expectation in disease's feature. Sixth, according to analysis based on the survey, patients would join again a hospital when they get satisfaction from the medical care and also they want to come again regarding to doctor's capability. For example, when doctors are old, have a good carrier and much experience, patients would come again. As seen from the above, consumers are usually satisfied with the medical treatment more than that they expected before. They would intend to use again when they get satisfaction from the medical care provided at a hospital. Patients and consumers highly expect good attitude as well as capacity from medical doctors and they are also generally satisfied with those things. Therefore, in order to increase the degree of consumer's satisfaction and their intention to come again, the hospital staff would have to commit themselves to achieve high quality service continuously and would have to make an effort to offer the finest quality service.

• Journal of Global Scholars of Marketing Science
• /
• v.11
• /
• pp.147-163
• /
• 2003

Recently it is necessary that Hotel business introduce Information Technology to enhance competitive advantage and cope with changeable business promptly in management. Thus in an effort of using Information Technology strategically, Many Hotel business tries to install CRM system (Customer Relationship Management). This study tries to analyze the effects on customer performance by installing CRM system if it is in charge of major strategic system, it can get successful customer performance. I hypothesize to resolve the problem, and search preceding study results concerning the elements of CRM Installment and Customer performance. The survey was taken to emplyees in the field of CRM installment in Luxury hotel to test the hypothesis. To summarize the results, first, CRM intallment affects CRM customer performance. in short, systematic feature, management environment, information intention, and technologic element affect it. Through this study, facing the limitation and future study are below. first, additional parameter should be considered though I reviewed the elements affecting CRM customer performance by searching and abstracting preceding studies. Second, There are lack of preceding studies because it has passed only a couple of years since Korean businesses deal with CRM system and there is the limitation to compare this result with others due to few empirical analysisses. Especilly, I can hardly find the preceding study concerning hotel industry but tries to search preceding parameter as to the customer performance of CRM system. Until now, It is needed to continual study its measurement later. I believe that more specific study and precise theoretical test be performed and they deal with current CRM system installment and facing problems in all of the korean hotels.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.25 no.5B
• /
• pp.942-955
• /
• 2000

This paper proposes enhanced designs of global service logic and information flow for the mass-call based IN service, which increase call completion rates and optimize the resource in joint wired and wireless networks. In order to hanve this logic implemented, we design a Dynamic Queue Manager(DQM) applied to the call queuing service feature in the Service Control Point(SCP). In order to apply this logic to wireless service subscribers as well as wired service subscribers, the service registration flags between the Home Location Register(HLR) and the SCP are managed to notify the DQM of the corresponding service subscribers’ mobility. Hence, we present a dynamic queue management mechanism, which dynamically manages the service group and the queue size based on M/M/c/K queueing model as the wireless subscribers roam the service groups due to their mobility characteristics. In order to determine the queue size allocated by the DQM, we simulator and analyze the relationship between the number of the subscriber’s terminals and the drop rate by considering the service increment rate. The appropriate waiting time in the queue as required is simulated according to the above relationship. Moreover, we design and implement the DQM that includes internal service logic interacting with SIBs(Service Independent building Blocks) and its data structure.

• Korean Journal of Environmental Agriculture
• /
• v.22 no.1
• /
• pp.53-59
• /
• 2003

The present study was conducted to analyze the vegetational difference in fellowed fields at the different seral stages. Plant species were surveyed on the fields having different cropping history, soil moisture conditions, and the duration of set-aside. Effects of soil moisture condition and fellowing duration on the vegetational profiles of fellowed fields in the course of different seral stages. In the fields fellowed for more than three years, a conspicuous feature of vegetation was the dominance of perennial species, which was less dominant in the fields fellowed for less than 3 years. The floristic composition of fellowed fields was dependent on the soil moisture condition. However, the influence of fallow-history on vegetational composition was less than that of the soil moisture conditions. The dominant species occurred in fellowed upland (dry) fields were changed from Glycine soja, Persicaria thunbergiana, and Artemisia princeps at 2-year-fellowed to Persicaria thunbergiana, Miscanthus sinensis, and Glycine soja at 6-year-fellowed. In wet fellowed paddy fields, annual Mosla punctulata, Ambrosia artemisiifolia, and Setaria viridis, the dominant species at 3-year-fellowed, were substituted by perennial Miscanthus sinensis, Aster pilosus, and Hemarthria sibirica at 7-year-fellowed. When the succession continued for 11 years in wet fields, the vegetation was characterized by the domination of perennials such as Phragmites communis, Zizania latifolia, and Typha orientalis. It was suggested that the soil moisture condition was a strong determinant of the dominant species on early seral conditions. In the fellowed paddy fields, the species diversity was relatively higher in the fields set-asided as wet condition compared to the fields fellowed as dry condition.

• Journal of KIISE:Computing Practices and Letters
• /
• v.6 no.4
• /
• pp.421-429
• /
• 2000

In present, ATM integrated switching system has been developed to a mixed modules that complexed switching system including maintenance, operation based on B-ISDN/LAN service and plug-in module, , which runs on workstation computer system. Meanwhile, workstation has HMI operation system feature including file system management, time management, graphic processing, TMN agent function. The workstation has communicated with between ATM switching module and clients. This computer system architecture has much burden messages communication among processes or processor. These messages communication consume system resources which are socket, message queue, IO device files, regular files, and so on. Therefore, in this paper we proposed new performance modeling with this system architecture. We will analyze the system bottleneck and improve system performance. In addition, in the future, the system has many additional features should be migrated to workstation system, we need previously to evaluate system bottleneck and redesign it. In performance model, we use queueing network model and the simulation package is used PDQ and C-program.

• Asian Pacific Journal of Cancer Prevention
• /
• v.15 no.9
• /
• pp.3927-3932
• /
• 2014

Background: Epidermal growth factor receptor (EGFR) mutations and echinoderm microtubule associated protein like 4-anaplastic lymphoma kinase (EML4-ALK) define specific molecular subsets of lung adenocarcinomas with distinct clinical features. Our purpose was to analyze clinical features and prognostic value of EGFR gene mutations and the EML4-ALK fusion gene in lung adenocarcinoma. Patients and Methods: EGFR gene mutations and the EML4-ALK fusion gene were detected in 92 lung adenocarcinoma patients in China. Tumor marker levels before first treatment were measured by electrochemiluminescence immunoassay. Results: EGFR mutations were found in 40.2% (37/92) of lung adenocarcinoma patients, being identified at high frequencies in never-smokers (48.3% vs. 26.5% in smokers; P=0.040) and in patients with abnormal serum carcinoembryonic antigen (CEA) levels before the initial treatment (58.3% vs. 28.6%, P=0.004). Multivariate analysis revealed that a higher serum CEA level before the initial treatment was independently associated with EGFR gene mutations (95%CI: 1.476~11.343, P=0.007). We also identified 8 patients who harbored the EML4-ALK fusion gene (8.7%, 8/92). In concordance with previous reports, younger age was a clinical feature for these (P=0.008). Seven of the positive cases were never smokers, and no coexistence with EGFR mutation was discovered. In addition, the frequency of the EML4-ALK fusion gene among patients with a serum CEA concentration below 5ng/ml seemed to be higher than patients with a concentration over 5ng/ml (P=0.021). No significant difference was observed for time to progression and overall survival between EML4-ALK-positive group and EML4-ALK-negative group or between patients with and without an EGFR mutation. Conclusions: The serum CEA level before the initial treatment may be helpful in screening population for EGFR mutations or EML4-ALK fusion gene presence in lung adenocarcinoma patients.

• Polymer(Korea)
• /
• v.39 no.1
• /
• pp.33-39
• /
• 2015

Olmesartan affiliated to biopharmaceutics classification system class 2 is a poorly water soluble drug. For this reason, olmesartan showed a low bioavailability and a lot of difficulties in the process of designing the pharmaceutical formulation. We prepared the solid dispersions of olmesartan. We confirmed the dissolution rate of drug which was prepared by manufacturing. The pharmaceutical formulation of solid dispersions was designed by using PVP as water soluble polymer. We analyzed morphological feature of solid dispersion by employing a scanning electron microscope. Then, the crystalline property of solid dispersion was confirmed through X-ray diffraction and differential scanning calorimeter. Also, the chemical change of solid dispersion was confirmed by the Fourier transform infrared spectroscopy. In vitro dissolution test was used to analyze the dissolution rate of solid dispersion. The prepared solid dissolution olmesartan confirmed the dissolution rate in the pH 1.2. It was compared with olmetec and improved dissolution rate through solid dispersion.