• Journal of KIISE:Software and Applications
• /
• v.30 no.11
• /
• pp.1054-1061
• /
• 2003

Q-Learning, a representative algorithm of reinforcement learning, experiences repeatedly until estimation values about all state-action pairs of state space converge and achieve optimal policies. When the state space is high dimensional or continuous, complex reinforcement learning tasks involve very large state space and suffer from storing all individual state values in a single table. We introduce Q-Map that is new function approximation method to get classified policies. As an agent learns on-line, Q-Map groups states of similar situations and adapts to new experiences repeatedly. State-action pairs necessary for fine control are treated in the form of rule. As a result of experiment in maze environment and mountain car problem, we can achieve classified knowledge and extract easily rules from Q-Map

• Journal of Intelligence and Information Systems
• /
• v.21 no.4
• /
• pp.111-131
• /
• 2015

There are only a handful number of research conducted on pattern analysis of corporate distress as compared with research for bankruptcy prediction. The few that exists mainly focus on audited firms because financial data collection is easier for these firms. But in reality, corporate financial distress is a far more common and critical phenomenon for non-audited firms which are mainly comprised of small and medium sized firms. The purpose of this paper is to classify non-audited firms under distress according to their financial ratio using data mining; Self-Organizing Map (SOM). SOM is a type of artificial neural network that is trained using unsupervised learning to produce a lower dimensional discretized representation of the input space of the training samples, called a map. SOM is different from other artificial neural networks as it applies competitive learning as opposed to error-correction learning such as backpropagation with gradient descent, and in the sense that it uses a neighborhood function to preserve the topological properties of the input space. It is one of the popular and successful clustering algorithm. In this study, we classify types of financial distress firms, specially, non-audited firms. In the empirical test, we collect 10 financial ratios of 100 non-audited firms under distress in 2004 for the previous two years (2002 and 2003). Using these financial ratios and the SOM algorithm, five distinct patterns were distinguished. In pattern 1, financial distress was very serious in almost all financial ratios. 12% of the firms are included in these patterns. In pattern 2, financial distress was weak in almost financial ratios. 14% of the firms are included in pattern 2. In pattern 3, growth ratio was the worst among all patterns. It is speculated that the firms of this pattern may be under distress due to severe competition in their industries. Approximately 30% of the firms fell into this group. In pattern 4, the growth ratio was higher than any other pattern but the cash ratio and profitability ratio were not at the level of the growth ratio. It is concluded that the firms of this pattern were under distress in pursuit of expanding their business. About 25% of the firms were in this pattern. Last, pattern 5 encompassed very solvent firms. Perhaps firms of this pattern were distressed due to a bad short-term strategic decision or due to problems with the enterpriser of the firms. Approximately 18% of the firms were under this pattern. This study has the academic and empirical contribution. In the perspectives of the academic contribution, non-audited companies that tend to be easily bankrupt and have the unstructured or easily manipulated financial data are classified by the data mining technology (Self-Organizing Map) rather than big sized audited firms that have the well prepared and reliable financial data. In the perspectives of the empirical one, even though the financial data of the non-audited firms are conducted to analyze, it is useful for find out the first order symptom of financial distress, which makes us to forecast the prediction of bankruptcy of the firms and to manage the early warning and alert signal. These are the academic and empirical contribution of this study. The limitation of this research is to analyze only 100 corporates due to the difficulty of collecting the financial data of the non-audited firms, which make us to be hard to proceed to the analysis by the category or size difference. Also, non-financial qualitative data is crucial for the analysis of bankruptcy. Thus, the non-financial qualitative factor is taken into account for the next study. This study sheds some light on the non-audited small and medium sized firms' distress prediction in the future.

• Atmosphere
• /
• v.19 no.2
• /
• pp.169-182
• /
• 2009

To improve the performance of climate and numerical models, concerns on the land-atmosphere schemes are steadily increased in recent years. For the realistic calculation of land-atmosphere interaction, a land surface information of high quality is strongly required. In this study, a new land cover map over the Korean peninsula was developed using MODIS (MODerate resolution Imaging Spectroradiometer) data. The seven phenological data set (maximum, minimum, amplitude, average, growing period, growing and shedding rate) derived from 15-day normalized difference vegetation index (NDVI) were used as a basic input data. The ISOData (Iterative Self-Organizing Data Analysis), a kind of unsupervised non-hierarchical clustering method, was applied to the seven phenological data set. After the clustering, assignment of land cover type to the each cluster was performed according to the phenological characteristics of each land cover defined by USGS (US. Geological Survey). Most of the Korean peninsula are occupied by deciduous broadleaf forest (46.5%), mixed forest (15.6%), and dryland crop (13%). Whereas, the dominant land cover types are very diverse in South-Korea: evergreen needleleaf forest (29.9%), mixed forest (26.6%), deciduous broadleaf forest (16.2%), irrigated crop (12.6%), and dryland crop (10.7%). The 38 in-situ observation data-base over South-Korea, Environment Geographic Information System and Google-earth are used in the validation of the new land cover map. In general, the new land cover map over the Korean peninsula seems to be better classified compared to the USGS land cover map, especially for the Savanna in the USGS land cover map.

• Journal of KIISE:Software and Applications
• /
• v.31 no.1
• /
• pp.61-70
• /
• 2004

It is difficult to find an appropriate web site because exponentially growing web contains millions of web documents. Personalization of web search can be realized by recommending proper web sites using user profile but more efficient method is needed for estimating preference because user's evaluation on web contents presents many aspects of his characteristics. As user profile has a property of non-linearity, estimation by classifier is needed and combination of classifiers is necessary to anticipate diverse properties. Structure adaptive self-organizing map (SASOM) that is suitable for Pattern classification and visualization is an enhanced model of SOM and might be useful for web mining. Fuzzy integral is a combination method using classifiers' relevance that is defined subjectively. In this paper, estimation of user profile is conducted by using ensemble of SASOM's teamed independently based on fuzzy integral and evaluated by Syskill ＆ Webert UCI benchmark data. Experimental results show that the proposed method performs better than previous naive Bayes classifier as well as voting of SASOM's.

• Journal of the Korean housing association
• /
• v.23 no.4
• /
• pp.59-67
• /
• 2012

Developing the modern design of Hanok and providing support for the commercialization model development in recent years propelled by the New Hanok Support Strategies of the central government in conjunction with the New Hanok revitalization related projects reflecting local goverments. New Hanok revitalization, the rekindling and revaluing of human behaviors and interests in local goverments following the social and cultural changes of the past decades, has emeraged as an increasingly traditional area of concerning in New Hanok planning. In this paper we attempt to this discussion by describing recent projects in New Hanok revitalization in Jeollanam-do Province. Therefore, this study aims to examine the classification of compound knowledges based multidimensional relationship by using Self-Organizing Maps (SOM). SOM is an unsupervised learning neural network model for the analysis of high-dimensional input data. By using SOM, we were able to create a cluster map reflecting the characteristics of the New Hanok. In this case the pattern of the preference data was easily understood by visual analysis. Liking for compound knowledge deduced from this data was classified into 8 categories according to the compound knowledge properties of New Hanok. As a result, a systematic approach for analysis the characteristics of individual family and living environment of New Hanoks and 10 space usage patterns the changes in some aspects of New Hanok.

• Journal of Convergence for Information Technology
• /
• v.12 no.5
• /
• pp.100-106
• /
• 2022

A longitudinal study refers to a research method based on longitudinal data repeatedly measured on the same object. Most of the longitudinal analysis methods are suitable for prediction or inference, and are often not suitable for use in exploratory study. In this study, an exploratory method to analyze longitudinal data is presented, which is to find the longitudinal trajectory after determining the best number of clusters by clustering longitudinal data using self-organizing map technique. The proposed methodology was applied to the longitudinal data of the Employment Information Service, and a total of 2,610 samples were analyzed. As a result of applying the methodology to the actual data applied, time-series clustering results were obtained for each panel. This indicates that it is more effective to cluster longitudinal data in advance and perform multilevel longitudinal analysis.

• Proceedings of the Korean Society of Computer Information Conference
• /
• 2011.01a
• /
• pp.137-140
• /
• 2011

시설의 안전성 평가에 대한 연구는 안전성에 영향을 주는 데이터를 정량화하여 획일적인 자동 수행하는 안전관리가 주를 이루고 있다. 이와 달리 자율수행은 수집 된 상황 정보나 상태 데이터를 이용하여 안전성을 예측하고 사고 위험성을 경보하여 사고를 예방 할 수 있다. 본 연구에서는 다양한 시설물 중에서 가스배관의 부식에 대한 판단을 위해서 신경망의 대표적 비지도학습인 자기조직화지도를 적용한다. SOM의 적용에서는 주변효과를 보완하기 위해서 관계적관점지도로 맵을 구성한다. 학습 할 데이터는 가스배관의 방식전위이다. 배관의 부식상태를 확인하기 위하여 수집 된 데이터인 방식전위에는 부식에 대한 위험요인이 내재되어 있다. 학습 후 새로운 데이터가 입력되면 각 상태 군집의 중심뉴런과 맵핑된 뉴런의 유사도를 측정하여 배관의 부식상태를 결정한다. 제안 된 방법으로 판단 된 결과를 기존에 사람이 판단한 결과와 비교하여 검증한다. 이를 통해 배관의 부식상태를 자율적이고 신속하게 판단하여 지능화 된 가스배관 관리로 활용한다.

• Asian Pacific Journal of Cancer Prevention
• /
• v.14 no.1
• /
• pp.111-116
• /
• 2013

Background: Cigarette smoking is the major risk factor for development of lung cancer. Identification of effects of tobacco on airway gene expression may provide insight into the causes. This research aimed to compare gene expression of large airway epithelium cells in normal smokers (n=13) and non-smokers (n=9) in order to find genes which discriminate the two groups and assess cigarette smoking effects on large airway epithelium cells.Materials and Methods: Genes discriminating smokers from non-smokers were identified by applying a neural network clustering method, growing self-organizing maps (GSOM), to microarray data according to class discrimination scores. An index was computed based on differentiation between each mean of gene expression in the two groups. This clustering approach provided the possibility of comparing thousands of genes simultaneously. Results: The applied approach compared the mean of 7,129 genes in smokers and non-smokers simultaneously and classified the genes of large airway epithelium cells which had differently expressed in smokers comparing with non-smokers. Seven genes were identified which had the highest different expression in smokers compared with the non-smokers group: NQO1, H19, ALDH3A1, AKR1C1, ABHD2, GPX2 and ADH7. Most (NQO1, ALDH3A1, AKR1C1, H19 and GPX2) are known to be clinically notable in lung cancer studies. Furthermore, statistical discriminate analysis showed that these genes could classify samples in smokers and non-smokers correctly with 100% accuracy. With the performed GSOM map, other nodes with high average discriminate scores included genes with alterations strongly related to the lung cancer such as AKR1C3, CYP1B1, UCHL1 and AKR1B10. Conclusions: This clustering by comparing expression of thousands of genes at the same time revealed alteration in normal smokers. Most of the identified genes were strongly relevant to lung cancer in the existing literature. The genes may be utilized to identify smokers with increased risk for lung cancer. A large sample study is now recommended to determine relations between the genes ABHD2 and ADH7 and smoking.

• Proceedings of the Korean Information Science Society Conference
• /
• 1999.10b
• /
• pp.75-77
• /
• 1999

정보추천 시스템은 사용자가 어떤 정보를 선호하는지를 식별함으로써 산재한 정보 중에서 적절한 정보만을 제공하는 것을 목표로 한다. 이러한 정보추천 시스템에서 사용되는 정보여과 기술에는 내용기반 여과와 협력적 여과가 있다. 기존의 협력적 정보여과 기술은 선호도를 적게 제시한 사용자에게 정보를 추천하기 어렵고, 동일한 상품 정보에 대해서 사용자의 평가가 없을 경우 사용자간의 유사성을 판단하기 어려운 단점이 있다. 본 논문은 SVD (Singular Value Decomposition)를 통해 사용자 프로파일을 정량화함으로써 사용자 선호도 행렬로부터 숨어있는 의미정보를 추출하여 동일한 정보에 대해 선호도를 평가해야 한다는 단점을 극복한다. 이때, 사용자 프로파일 벡터를 비감독 학습 알고리즘인 SOM (Self0Organizing Map)으로 클러스터링하여 사용자를 분류하고, 정보추천은 사용자 그룹간에서 이루어지며 Pearson correlation 알고리즘을 이용한다. 기존의 방법과 비교한 결과, 제안한 방법이 새로운 사용자에 대해서도 적절한 정보를 추천할 수 있음을 볼 수 있었다.

• Proceedings of the Korean Information Science Society Conference
• /
• 1999.10b
• /
• pp.333-335
• /
• 1999

일반적으로 자기구성 지도에서는 구조가 초기에 결정되어 학습이 끝날때까지 변하기 않기 때문에 각 문제에 대한 구조를 반복된 실험을 통해서 최적화시켜야 한다. 그러나, 지도의 구조가 학습중에 적절하게 변경된다면, 해당 문제에 가장 알맞은 구조의 지도를 생성할 수 있을 것이다. 이 논문에서는 기존의 적응형 자기 구성 지도의 비교사 학습방법에 LVQ 알고리즘을 이용한 교사 학습방법을 결합한 구조 적응형 자기 구성 지도 모델을 제안한다. 이 방법은 일반적인 자기구성 지도 알고리즘보다 작은 수의 노드를 가지고 높은 성능을 보일 뿐만 아니라, 자기 구성 지도의 특성인 위상 보존도 잘 이루어진다. 오프라인 필기 숫자 데이터로 실험한 결과, 제안한 방법이 유용함을 알 수 있었다.