• Journal of Intelligence and Information Systems
• /
• v.25 no.4
• /
• pp.53-65
• /
• 2019

Object tracking is one of important steps to achieve video-based surveillance systems. Object tracking is considered as an essential task similar to object detection and recognition. In order to perform object tracking, various machine learning methods (e.g., least-squares, perceptron and support vector machine) can be applied for different designs of tracking systems. In general, generative methods (e.g., principal component analysis) were utilized due to its simplicity and effectiveness. However, the generative methods were only focused on modeling the target object. Due to this limitation, discriminative methods (e.g., binary classification) were adopted to distinguish the target object and the background. Among the machine learning methods for binary classification, total error rate minimization can be used as one of successful machine learning methods for binary classification. The total error rate minimization can achieve a global minimum due to a quadratic approximation to a step function while other methods (e.g., support vector machine) seek local minima using nonlinear functions (e.g., hinge loss function). Due to this quadratic approximation, the total error rate minimization could obtain appropriate properties in solving optimization problems for binary classification. However, this total error rate minimization was based on a batch mode setting. The batch mode setting can be limited to several applications under offline learning. Due to limited computing resources, offline learning could not handle large scale data sets. Compared to offline learning, online learning can update its solution without storing all training samples in learning process. Due to increment of large scale data sets, online learning becomes one of essential properties for various applications. Since object tracking needs to handle data samples in real time, online learning based total error rate minimization methods are necessary to efficiently address object tracking problems. Due to the need of the online learning, an online learning based total error rate minimization method was developed. However, an approximately reweighted technique was developed. Although the approximation technique is utilized, this online version of the total error rate minimization could achieve good performances in biometric applications. However, this method is assumed that the total error rate minimization can be asymptotically achieved when only the number of training samples is infinite. Although there is the assumption to achieve the total error rate minimization, the approximation issue can continuously accumulate learning errors according to increment of training samples. Due to this reason, the approximated online learning solution can then lead a wrong solution. The wrong solution can make significant errors when it is applied to surveillance systems. In this paper, we propose an exactly reweighted technique to recursively update the solution of the total error rate minimization in online learning manner. Compared to the approximately reweighted online total error rate minimization, an exactly reweighted online total error rate minimization is achieved. The proposed exact online learning method based on the total error rate minimization is then applied to object tracking problems. In our object tracking system, particle filtering is adopted. In particle filtering, our observation model is consisted of both generative and discriminative methods to leverage the advantages between generative and discriminative properties. In our experiments, our proposed object tracking system achieves promising performances on 8 public video sequences over competing object tracking systems. The paired t-test is also reported to evaluate its quality of the results. Our proposed online learning method can be extended under the deep learning architecture which can cover the shallow and deep networks. Moreover, online learning methods, that need the exact reweighting process, can use our proposed reweighting technique. In addition to object tracking, the proposed online learning method can be easily applied to object detection and recognition. Therefore, our proposed methods can contribute to online learning community and object tracking, detection and recognition communities.

• Journal of Intelligence and Information Systems
• /
• v.24 no.4
• /
• pp.33-49
• /
• 2018

The impact of TV Drama success on TV Rating and the channel promotion effectiveness is very high. The cultural and business impact has been also demonstrated through the Korean Wave. Therefore, the early prediction of the blockbuster success of TV Drama is very important from the strategic perspective of the media industry. Previous studies have tried to predict the audience ratings and success of drama based on various methods. However, most of the studies have made simple predictions using intuitive methods such as the main actor and time zone. These studies have limitations in predicting. In this study, we propose a model for predicting the popularity of drama by analyzing the customer's viewing pattern based on various theories. This is not only a theoretical contribution but also has a contribution from the practical point of view that can be used in actual broadcasting companies. In this study, we collected data of 280 TV mini-series dramas, broadcasted over the terrestrial channels for 10 years from 2003 to 2012. From the data, we selected the most highly ranked and the least highly ranked 45 TV drama and analyzed the viewing patterns of them by 11-step. The various assumptions and conditions for modeling are based on existing studies, or by the opinions of actual broadcasters and by data mining techniques. Then, we developed a prediction model by measuring the viewing-time distance (difference) using Euclidean and Correlation method, which is termed in our study similarity (the sum of distance). Through the similarity measure, we predicted the success of dramas from the viewer's initial viewing-time pattern distribution using 1~5 episodes. In order to confirm that the model is shaken according to the measurement method, various distance measurement methods were applied and the model was checked for its dryness. And when the model was established, we could make a more predictive model using a grid search. Furthermore, we classified the viewers who had watched TV drama more than 70% of the total airtime as the "passionate viewer" when a new drama is broadcasted. Then we compared the drama's passionate viewer percentage the most highly ranked and the least highly ranked dramas. So that we can determine the possibility of blockbuster TV mini-series. We find that the initial viewing-time pattern is the key factor for the prediction of blockbuster dramas. From our model, block-buster dramas were correctly classified with the 75.47% accuracy with the initial viewing-time pattern analysis. This paper shows high prediction rate while suggesting audience rating method different from existing ones. Currently, broadcasters rely heavily on some famous actors called so-called star systems, so they are in more severe competition than ever due to rising production costs of broadcasting programs, long-term recession, aggressive investment in comprehensive programming channels and large corporations. Everyone is in a financially difficult situation. The basic revenue model of these broadcasters is advertising, and the execution of advertising is based on audience rating as a basic index. In the drama, there is uncertainty in the drama market that it is difficult to forecast the demand due to the nature of the commodity, while the drama market has a high financial contribution in the success of various contents of the broadcasting company. Therefore, to minimize the risk of failure. Thus, by analyzing the distribution of the first-time viewing time, it can be a practical help to establish a response strategy (organization/ marketing/story change, etc.) of the related company. Also, in this paper, we found that the behavior of the audience is crucial to the success of the program. In this paper, we define TV viewing as a measure of how enthusiastically watching TV is watched. We can predict the success of the program successfully by calculating the loyalty of the customer with the hot blood. This way of calculating loyalty can also be used to calculate loyalty to various platforms. It can also be used for marketing programs such as highlights, script previews, making movies, characters, games, and other marketing projects.

• Journal of Intelligence and Information Systems
• /
• v.25 no.3
• /
• pp.239-251
• /
• 2019

Statistical methods such as moving averages, Kalman filtering, exponential smoothing, regression analysis, and ARIMA (autoregressive integrated moving average) have been used for stock market predictions. However, these statistical methods have not produced superior performances. In recent years, machine learning techniques have been widely used in stock market predictions, including artificial neural network, SVM, and genetic algorithm. In particular, a case-based reasoning method, known as k-nearest neighbor is also widely used for stock price prediction. Case based reasoning retrieves several similar cases from previous cases when a new problem occurs, and combines the class labels of similar cases to create a classification for the new problem. However, case based reasoning has some problems. First, case based reasoning has a tendency to search for a fixed number of neighbors in the observation space and always selects the same number of neighbors rather than the best similar neighbors for the target case. So, case based reasoning may have to take into account more cases even when there are fewer cases applicable depending on the subject. Second, case based reasoning may select neighbors that are far away from the target case. Thus, case based reasoning does not guarantee an optimal pseudo-neighborhood for various target cases, and the predictability can be degraded due to a deviation from the desired similar neighbor. This paper examines how the size of learning data affects stock price predictability through k-nearest neighbor and compares the predictability of k-nearest neighbor with the random walk model according to the size of the learning data and the number of neighbors. In this study, Samsung electronics stock prices were predicted by dividing the learning dataset into two types. For the prediction of next day's closing price, we used four variables: opening value, daily high, daily low, and daily close. In the first experiment, data from January 1, 2000 to December 31, 2017 were used for the learning process. In the second experiment, data from January 1, 2015 to December 31, 2017 were used for the learning process. The test data is from January 1, 2018 to August 31, 2018 for both experiments. We compared the performance of k-NN with the random walk model using the two learning dataset. The mean absolute percentage error (MAPE) was 1.3497 for the random walk model and 1.3570 for the k-NN for the first experiment when the learning data was small. However, the mean absolute percentage error (MAPE) for the random walk model was 1.3497 and the k-NN was 1.2928 for the second experiment when the learning data was large. These results show that the prediction power when more learning data are used is higher than when less learning data are used. Also, this paper shows that k-NN generally produces a better predictive power than random walk model for larger learning datasets and does not when the learning dataset is relatively small. Future studies need to consider macroeconomic variables related to stock price forecasting including opening price, low price, high price, and closing price. Also, to produce better results, it is recommended that the k-nearest neighbor needs to find nearest neighbors using the second step filtering method considering fundamental economic variables as well as a sufficient amount of learning data.

• Journal of Intelligence and Information Systems
• /
• v.25 no.3
• /
• pp.43-62
• /
• 2019

At one time, the anomaly detection sector dominated the method of determining whether there was an abnormality based on the statistics derived from specific data. This methodology was possible because the dimension of the data was simple in the past, so the classical statistical method could work effectively. However, as the characteristics of data have changed complexly in the era of big data, it has become more difficult to accurately analyze and predict the data that occurs throughout the industry in the conventional way. Therefore, SVM and Decision Tree based supervised learning algorithms were used. However, there is peculiarity that supervised learning based model can only accurately predict the test data, when the number of classes is equal to the number of normal classes and most of the data generated in the industry has unbalanced data class. Therefore, the predicted results are not always valid when supervised learning model is applied. In order to overcome these drawbacks, many studies now use the unsupervised learning-based model that is not influenced by class distribution, such as autoencoder or generative adversarial networks. In this paper, we propose a method to detect anomalies using generative adversarial networks. AnoGAN, introduced in the study of Thomas et al (2017), is a classification model that performs abnormal detection of medical images. It was composed of a Convolution Neural Net and was used in the field of detection. On the other hand, sequencing data abnormality detection using generative adversarial network is a lack of research papers compared to image data. Of course, in Li et al (2018), a study by Li et al (LSTM), a type of recurrent neural network, has proposed a model to classify the abnormities of numerical sequence data, but it has not been used for categorical sequence data, as well as feature matching method applied by salans et al.(2016). So it suggests that there are a number of studies to be tried on in the ideal classification of sequence data through a generative adversarial Network. In order to learn the sequence data, the structure of the generative adversarial networks is composed of LSTM, and the 2 stacked-LSTM of the generator is composed of 32-dim hidden unit layers and 64-dim hidden unit layers. The LSTM of the discriminator consists of 64-dim hidden unit layer were used. In the process of deriving abnormal scores from existing paper of Anomaly Detection for Sequence data, entropy values of probability of actual data are used in the process of deriving abnormal scores. but in this paper, as mentioned earlier, abnormal scores have been derived by using feature matching techniques. In addition, the process of optimizing latent variables was designed with LSTM to improve model performance. The modified form of generative adversarial model was more accurate in all experiments than the autoencoder in terms of precision and was approximately 7% higher in accuracy. In terms of Robustness, Generative adversarial networks also performed better than autoencoder. Because generative adversarial networks can learn data distribution from real categorical sequence data, Unaffected by a single normal data. But autoencoder is not. Result of Robustness test showed that he accuracy of the autocoder was 92%, the accuracy of the hostile neural network was 96%, and in terms of sensitivity, the autocoder was 40% and the hostile neural network was 51%. In this paper, experiments have also been conducted to show how much performance changes due to differences in the optimization structure of potential variables. As a result, the level of 1% was improved in terms of sensitivity. These results suggest that it presented a new perspective on optimizing latent variable that were relatively insignificant.

• Journal of Intelligence and Information Systems
• /
• v.25 no.3
• /
• pp.63-88
• /
• 2019

The goal of this paper is to investigate changes in North Korea's domestic and foreign policies through automated text analysis over North Korea represented in South Korean mass media. Based on that data, we then analyze the status of text mining research, using a text mining technique to find the topics, methods, and trends of text mining research. We also investigate the characteristics and method of analysis of the text mining techniques, confirmed by analysis of the data. In this study, R program was used to apply the text mining technique. R program is free software for statistical computing and graphics. Also, Text mining methods allow to highlight the most frequently used keywords in a paragraph of texts. One can create a word cloud, also referred as text cloud or tag cloud. This study proposes a procedure to find meaningful tendencies based on a combination of word cloud, and co-occurrence networks. This study aims to more objectively explore the images of North Korea represented in South Korean newspapers by quantitatively reviewing the patterns of language use related to North Korea from 2016. 11. 1 to 2019. 5. 23 newspaper big data. In this study, we divided into three periods considering recent inter - Korean relations. Before January 1, 2018, it was set as a Before Phase of Peace Building. From January 1, 2018 to February 24, 2019, we have set up a Peace Building Phase. The New Year's message of Kim Jong-un and the Olympics of Pyeong Chang formed an atmosphere of peace on the Korean peninsula. After the Hanoi Pease summit, the third period was the silence of the relationship between North Korea and the United States. Therefore, it was called Depression Phase of Peace Building. This study analyzes news articles related to North Korea of the Korea Press Foundation database(www.bigkinds.or.kr) through text mining, to investigate characteristics of the Kim Jong-un regime's South Korea policy and unification discourse. The main results of this study show that trends in the North Korean national policy agenda can be discovered based on clustering and visualization algorithms. In particular, it examines the changes in the international circumstances, domestic conflicts, the living conditions of North Korea, the South's Aid project for the North, the conflicts of the two Koreas, North Korean nuclear issue, and the North Korean refugee problem through the co-occurrence word analysis. It also offers an analysis of South Korean mentality toward North Korea in terms of the semantic prosody. In the Before Phase of Peace Building, the results of the analysis showed the order of 'Missiles', 'North Korea Nuclear', 'Diplomacy', 'Unification', and ' South-North Korean'. The results of Peace Building Phase are extracted the order of 'Panmunjom', 'Unification', 'North Korea Nuclear', 'Diplomacy', and 'Military'. The results of Depression Phase of Peace Building derived the order of 'North Korea Nuclear', 'North and South Korea', 'Missile', 'State Department', and 'International'. There are 16 words adopted in all three periods. The order is as follows: 'missile', 'North Korea Nuclear', 'Diplomacy', 'Unification', 'North and South Korea', 'Military', 'Kaesong Industrial Complex', 'Defense', 'Sanctions', 'Denuclearization', 'Peace', 'Exchange and Cooperation', and 'South Korea'. We expect that the results of this study will contribute to analyze the trends of news content of North Korea associated with North Korea's provocations. And future research on North Korean trends will be conducted based on the results of this study. We will continue to study the model development for North Korea risk measurement that can anticipate and respond to North Korea's behavior in advance. We expect that the text mining analysis method and the scientific data analysis technique will be applied to North Korea and unification research field. Through these academic studies, I hope to see a lot of studies that make important contributions to the nation.

• Journal of Intelligence and Information Systems
• /
• v.25 no.1
• /
• pp.127-137
• /
• 2019

Recidivism prediction has been a subject of constant research by experts since the early 1970s. But it has become more important as committed crimes by recidivist steadily increase. Especially, in the 1990s, after the US and Canada adopted the 'Recidivism Risk Assessment Report' as a decisive criterion during trial and parole screening, research on recidivism prediction became more active. And in the same period, empirical studies on 'Recidivism Factors' were started even at Korea. Even though most recidivism prediction studies have so far focused on factors of recidivism or the accuracy of recidivism prediction, it is important to minimize the prediction misclassification cost, because recidivism prediction has an asymmetric error cost structure. In general, the cost of misrecognizing people who do not cause recidivism to cause recidivism is lower than the cost of incorrectly classifying people who would cause recidivism. Because the former increases only the additional monitoring costs, while the latter increases the amount of social, and economic costs. Therefore, in this paper, we propose an XGBoost(eXtream Gradient Boosting; XGB) based recidivism prediction model considering asymmetric error cost. In the first step of the model, XGB, being recognized as high performance ensemble method in the field of data mining, was applied. And the results of XGB were compared with various prediction models such as LOGIT(logistic regression analysis), DT(decision trees), ANN(artificial neural networks), and SVM(support vector machines). In the next step, the threshold is optimized to minimize the total misclassification cost, which is the weighted average of FNE(False Negative Error) and FPE(False Positive Error). To verify the usefulness of the model, the model was applied to a real recidivism prediction dataset. As a result, it was confirmed that the XGB model not only showed better prediction accuracy than other prediction models but also reduced the cost of misclassification most effectively.

• Journal of Intelligence and Information Systems
• /
• v.27 no.4
• /
• pp.49-71
• /
• 2021

Users who intend to utilize knowledge to actively solve given problems proceed their jobs with cross- and sequential exploration of associated knowledge related each other in terms of certain criteria, such as content relevance. A knowledge map is the diagram or taxonomy overviewing status of currently managed knowledge in a knowledge-base, and supports users' knowledge exploration based on certain relationships between knowledge. A knowledge map, therefore, must be expressed in a networked form by linking related knowledge based on certain types of relationships, and should be implemented by deploying proper technologies or tools specialized in defining and inferring them. To meet this end, this study suggests a methodology for developing the knowledge graph-based knowledge map using the Graph DB known to exhibit proper functionality in expressing and inferring relationships between entities and their relationships stored in a knowledge-base. Procedures of the proposed methodology are modeling graph data, creating nodes, properties, relationships, and composing knowledge networks by combining identified links between knowledge. Among various Graph DBs, the Neo4j is used in this study for its high credibility and applicability through wide and various application cases. To examine the validity of the proposed methodology, a knowledge graph-based knowledge map is implemented deploying the Graph DB, and a performance comparison test is performed, by applying previous research's data to check whether this study's knowledge map can yield the same level of performance as the previous one did. Previous research's case is concerned with building a process-based knowledge map using the ontology technology, which identifies links between related knowledge based on the sequences of tasks producing or being activated by knowledge. In other words, since a task not only is activated by knowledge as an input but also produces knowledge as an output, input and output knowledge are linked as a flow by the task. Also since a business process is composed of affiliated tasks to fulfill the purpose of the process, the knowledge networks within a business process can be concluded by the sequences of the tasks composing the process. Therefore, using the Neo4j, considered process, task, and knowledge as well as the relationships among them are defined as nodes and relationships so that knowledge links can be identified based on the sequences of tasks. The resultant knowledge network by aggregating identified knowledge links is the knowledge map equipping functionality as a knowledge graph, and therefore its performance needs to be tested whether it meets the level of previous research's validation results. The performance test examines two aspects, the correctness of knowledge links and the possibility of inferring new types of knowledge: the former is examined using 7 questions, and the latter is checked by extracting two new-typed knowledge. As a result, the knowledge map constructed through the proposed methodology has showed the same level of performance as the previous one, and processed knowledge definition as well as knowledge relationship inference in a more efficient manner. Furthermore, comparing to the previous research's ontology-based approach, this study's Graph DB-based approach has also showed more beneficial functionality in intensively managing only the knowledge of interest, dynamically defining knowledge and relationships by reflecting various meanings from situations to purposes, agilely inferring knowledge and relationships through Cypher-based query, and easily creating a new relationship by aggregating existing ones, etc. This study's artifacts can be applied to implement the user-friendly function of knowledge exploration reflecting user's cognitive process toward associated knowledge, and can further underpin the development of an intelligent knowledge-base expanding autonomously through the discovery of new knowledge and their relationships by inference. This study, moreover than these, has an instant effect on implementing the networked knowledge map essential to satisfying contemporary users eagerly excavating the way to find proper knowledge to use.

• Journal of Intelligence and Information Systems
• /
• v.28 no.2
• /
• pp.191-206
• /
• 2022

Recently, it is a de-facto approach to utilize a pre-trained language model(PLM) to achieve the state-of-the-art performance for various natural language tasks(called downstream tasks) such as sentiment analysis and question answering. However, similar to any other machine learning method, PLM tends to depend on the data distribution seen during the training phase and shows worse performance on the unseen (Out-of-Distribution) domain. Due to the aforementioned reason, there have been many efforts to develop domain-specified PLM for various fields such as medical and legal industries. In this paper, we discuss the training of a finance domain-specified PLM for the Korean language and its applications. Our finance domain-specified PLM, KB-BERT, is trained on a carefully curated financial corpus that includes domain-specific documents such as financial reports. We provide extensive performance evaluation results on three natural language tasks, topic classification, sentiment analysis, and question answering. Compared to the state-of-the-art Korean PLM models such as KoELECTRA and KLUE-RoBERTa, KB-BERT shows comparable performance on general datasets based on common corpora like Wikipedia and news articles. Moreover, KB-BERT outperforms compared models on finance domain datasets that require finance-specific knowledge to solve given problems.

• Journal of Intelligence and Information Systems
• /
• v.28 no.3
• /
• pp.105-117
• /
• 2022

This study analyzes the effects between stock returns and interest rate spread, difference between long-term and short-term interest rate through the polynomial linear regression analysis. The existing research concentrated on the business forecast through the interest rate spread focusing on the US market. The previous studies verified the interest rate spread based on the leading indicators of business forecast by moderating the period of long-term/short-term interest rates and analyzing the degree of leading. After the 7th reform of composite indices of business indicators in Korea of 2006, the interest rate spread was included in the items of composing the business leading indicators, which is utilized till today. Nevertheless, there are a few research on stock returns of each industry and interest rate spread in domestic stock market. Therefore, this study analyzed the stock returns of each industry and interest rate spread targeting Korean stock market. This study selected the long-term/short-term interest rates with high causality through the regression analysis, and then understood the correlations with each leading period and industry. To overcome the limitation of the simple linear regression analysis, polynomial linear regression analysis is used, which raised explanatory power. As a result, the high causality was verified when using differences between returns of corporate bond(AA-) without guarantee for three years by leading six months and call rate returns as interest rate spread. In addition, analyzing the stock returns of each industry, the relation between the relevant interest rate spread and returns of the automobile industry was the closest. This study is significant in the aspect of verifying the causality of interest rate spread, business forecast, and stock returns in Korea. Even though it could be limited to forecast the stock price by using only the interest rate spread, it would be working as a strong factor when it is properly utilized with other various factors.

• Journal of Intelligence and Information Systems
• /
• v.28 no.2
• /
• pp.127-146
• /
• 2022

Recently, as word embedding has shown excellent performance in various tasks of deep learning-based natural language processing, researches on the advancement and application of word, sentence, and document embedding are being actively conducted. Among them, cross-language transfer, which enables semantic exchange between different languages, is growing simultaneously with the development of embedding models. Academia's interests in vector alignment are growing with the expectation that it can be applied to various embedding-based analysis. In particular, vector alignment is expected to be applied to mapping between specialized domains and generalized domains. In other words, it is expected that it will be possible to map the vocabulary of specialized fields such as R&D, medicine, and law into the space of the pre-trained language model learned with huge volume of general-purpose documents, or provide a clue for mapping vocabulary between mutually different specialized fields. However, since linear-based vector alignment which has been mainly studied in academia basically assumes statistical linearity, it tends to simplify the vector space. This essentially assumes that different types of vector spaces are geometrically similar, which yields a limitation that it causes inevitable distortion in the alignment process. To overcome this limitation, we propose a deep learning-based vector alignment methodology that effectively learns the nonlinearity of data. The proposed methodology consists of sequential learning of a skip-connected autoencoder and a regression model to align the specialized word embedding expressed in each space to the general embedding space. Finally, through the inference of the two trained models, the specialized vocabulary can be aligned in the general space. To verify the performance of the proposed methodology, an experiment was performed on a total of 77,578 documents in the field of 'health care' among national R&D tasks performed from 2011 to 2020. As a result, it was confirmed that the proposed methodology showed superior performance in terms of cosine similarity compared to the existing linear vector alignment.