• Journal of Information Processing Systems
• /
• v.16 no.3
• /
• pp.612-628
• /
• 2020

This paper presents a new approach for detecting abnormal behaviors in complex surveillance scenes where anomalies are subtle and difficult to distinguish due to the intricate correlations among multiple objects' behaviors. Specifically, a cascaded probabilistic topic model was put forward for learning the spatial context of local behavior and the temporal context of global behavior in two different stages. In the first stage of topic modeling, unlike the existing approaches using either optical flows or complete trajectories, spatio-temporal correlations between the trajectory fragments in video clips were modeled by the latent Dirichlet allocation (LDA) topic model based on Markov random fields to obtain the spatial context of local behavior in each video clip. The local behavior topic categories were then obtained by exploiting the spectral clustering algorithm. Based on the construction of a dictionary through the process of local behavior topic clustering, the second phase of the LDA topic model learns the correlations of global behaviors and temporal context. In particular, an abnormal behavior recognition method was developed based on the learned spatio-temporal context of behaviors. The specific identification method adopts a top-down strategy and consists of two stages: anomaly recognition of video clip and anomalous behavior recognition within each video clip. Evaluation was performed using the validity of spatio-temporal context learning for local behavior topics and abnormal behavior recognition. Furthermore, the performance of the proposed approach in abnormal behavior recognition improved effectively and significantly in complex surveillance scenes.

• IE interfaces
• /
• v.24 no.4
• /
• pp.447-456
• /
• 2011

Identifying the major moving patterns of shoppers' movements in the selling floor has been a longstanding issue in the retailing industry. With the advent of RFID technology, it has been easier to collect the moving data for a individual shopper's movement. Most of the previous studies used the traditional clustering technique to identify the major moving pattern of customers. However, in using clustering technique, due to the spatial constraint (aisle layout or other physical obstructions in the store), standard clustering methods are not feasible for moving data like shopping path should be adjusted for the analysis in advance, which is time-consuming and causes data distortion. To alleviate this problems, we propose a new approach to spatial pattern clustering based on longest common subsequence (LCSS). Experimental results using the real data obtained from a grocery in Seoul show that the proposed method performs well in finding the hot spot and dead spot as well as in finding the major path patterns of customer movements.

• The Journal of Society for e-Business Studies
• /
• v.27 no.2
• /
• pp.193-204
• /
• 2022

Smart card data is representative mobility data and can be used for policy development by analyzing public transportation usage behavior. This paper deals with the problem of classifying metro stations using metro usage patterns as one of these studies. Since the previous papers dealing with clustering of metro stations only considered traffic among usage behaviors, this paper proposes clustering considering traffic time as one of the complementary methods. Passengers at each station were classified into passengers arriving at work time, arriving at quitting time, leaving at work time, and leaving at quitting time, and then the estimated shape parameter was defined as the characteristic value of the station by modeling each transit time to Weibull distribution. And the characteristic vectors were clustered using the K-means clustering technique. As a result of the experiment, it was observed that station clustering considering pass time is not only similar to the clustering results of previous studies, but also enables more granular clustering.

• Journal of Preventive Medicine and Public Health
• /
• v.43 no.1
• /
• pp.73-83
• /
• 2010

Objectives: We investigated the clustering of selected lifestyle factors (cigarette smoking, heavy alcohol consumption, lack of physical exercise) and identified the population characteristics associated with increasing lifestyle risks. Methods: Data on lifestyle risk factors, sociodemographic characteristics, and history of chronic diseases were obtained from 7,694 individuals ${\geq}20$ years of age who participated in the 2005 Korea National Health and Nutrition Examination Survey (KNHANES). Clustering of lifestyle risks involved the observed prevalence of multiple risks and those expected from marginal exposure prevalence of the three selected risk factors. Prevalence odds ratio was adopted as a measurement of clustering. Multiple correspondence analysis, Kendall tau correlation, Man-Whitney analysis, and ordinal logistic regression analysis were conducted to identify variables increasing lifestyle risks. Results: In both men and women, increased lifestyle risks were associated with clustering of: (1) cigarette smoking and excessive alcohol consumption, and (2) smoking, excessive alcohol consumption, and lack of physical exercise. Patterns of clustering for physical exercise were different from those for cigarette smoking and alcohol consumption. The increased unhealthy clustering was found among men 20-64 years of age with mild or moderate stress, and among women 35-49 years of age who were never-married, with mild stress, and increased body mass index (>$30\;kg/m^2$). Conclusions: Addressing a lack of physical exercise considering individual characteristics including gender, age, employment activity, and stress levels should be a focus of health promotion efforts.

• The Journal of the Korea Contents Association
• /
• v.7 no.6
• /
• pp.22-30
• /
• 2007

Most real word systems such as world economy, stock market, and medical applications, contain a series of dynamic and complex phenomena. One of common methods to understand these systems is to build a model and analyze the behavior of the system. In this paper, we investigated methods for best clustering over time series data. As a first step for clustering, BIC (Bayesian Information Criterion) approximation is used to determine the number of clusters. A search technique to improve clustering efficiency is also suggested by analyzing the relationship between data size and BIC values. For clustering, two methods, model-based and similarity based methods, are analyzed and compared. A number of experiments have been performed to check its validity using real data(stock price). BIC approximation measure has been confirmed that it suggests best number of clusters through experiments provided that the number of data is relatively large. It is also confirmed that the model-based clustering produces more reliable clustering than similarity based ones.

• Journal of Korean Society of Industrial and Systems Engineering
• /
• v.40 no.4
• /
• pp.203-210
• /
• 2017

Data clustering is one of the most difficult and challenging problems and can be formally considered as a particular kind of NP-hard grouping problems. The K-means algorithm is one of the most popular and widely used clustering method because it is easy to implement and very efficient. However, it has high possibility to trap in local optimum and high variation of solutions with different initials for the large data set. Therefore, we need study efficient computational intelligence method to find the global optimal solution in data clustering problem within limited computational time. The objective of this paper is to propose a combined artificial bee colony (CABC) with K-means for initialization and finalization to find optimal solution that is effective on data clustering optimization problem. The artificial bee colony (ABC) is an algorithm motivated by the intelligent behavior exhibited by honeybees when searching for food. The performance of ABC is better than or similar to other population-based algorithms with the added advantage of employing fewer control parameters. Our proposed CABC method is able to provide near optimal solution within reasonable time to balance the converged and diversified searches. In this paper, the experiment and analysis of clustering problems demonstrate that CABC is a competitive approach comparing to previous partitioning approaches in satisfactory results with respect to solution quality. We validate the performance of CABC using Iris, Wine, Glass, Vowel, and Cloud UCI machine learning repository datasets comparing to previous studies by experiment and analysis. Our proposed KABCK (K-means+ABC+K-means) is better than ABCK (ABC+K-means), KABC (K-means+ABC), ABC, and K-means in our simulations.

• Journal of KIISE:Databases
• /
• v.35 no.1
• /
• pp.84-95
• /
• 2008

Wireless sensor networks have recently emerged as a platform for several applications. By deploying wireless sensor nodes and constructing a sensor network, we can remotely obtain information about the behavior, conditions, and positions of objects in a region. Since sensor nodes operate on batteries, energy-efficient mechanisms for gathering sensor data are indispensable to prolong the lifetime of a sensor network as long as possible. In this paper, we propose a novel clustering algorithm that distributes the energy consumption of a cluster head. First, we analyze the energy consumption if cluster heads and divide each cluster into a collection layer and a transmission layer according to their roles. Then, we elect a cluster head for each layer to distribute the energy consumption of single cluster head. In order to show the superiority of our clustering algorithm, we compare it with the existing clustering algorithm in terms of the lifetime of the sensor network. As a result, our experimental results show that the proposed clustering algorithm achieves about $10%{\sim}40%$ performance improvements over the existing clustering algorithms.

• Proceedings of the KIEE Conference
• /
• 1998.07g
• /
• pp.2186-2188
• /
• 1998

In this paper, Fundamental rules governing group intelligence "arrangement" behavior of multiple number of autonomous mobile robots are represented by a small number of fuzzy rules. Complex lifelike behavior is considered as local interactions between simple individuals under small number of fundamental rules. The fuzzy rules for arrangement are generated from clustering the input-output data obtained from the arrangement algorithm. Simulation shows the fuzzy rules successfully realizes fundamental rules of the flocking group behavior.

• Health Policy and Management
• /
• v.30 no.2
• /
• pp.199-210
• /
• 2020

Background: Based on the importance of ceasing smoking programs to control the regional disparity of smoking behavior in Korea, this study aims to reveal the variation of smoke rate and determinants of it for 229 provinces. An evaluation of the relative efficiency of the cease smoking program under the consideration of regional characteristics was followed. Methods: The main sources of data are the Korean Statistical Information Service and a national survey on the expenditure of public health centers. Multivariate regression is performed to figure the determinants of regional variation of smoking rate. Based on the result of the regression model, clustering analysis was conducted to group 229 regions by their characteristics. Three clusters were generated. Using data envelopment analysis (DEA), relative efficiency scores are calculated. Results from the pooled model which put 229 provinces in one model to score relative efficiency were compared with the cluster-separated model of each cluster. Results: First, the maximum variation of the smoking rate was 16.9%p. Second, sex ration, the proportion of the elder, and high risk drinking alcohol behavior have a significant role in the regional variation of smoking. Third, the population and proportion of the elder are the main variables for clustering. Fourth, dissimilarity on the results of relative efficiency was found between the pooled model and cluster-separated model, especially for cluster 2. Conclusion: This study figured regional variation of smoking rate and its determinants on the regional level. Unconformity of the DEA results between different models implies the issues on regional features when the regional evaluation performed especially on the programs of public health centers.

• Journal of Institute of Control, Robotics and Systems
• /
• v.13 no.6
• /
• pp.542-546
• /
• 2007

In the vision of ubiquitous computing environment, smart objects would communicate each other and provide many kinds of information about user and their surroundings in the home. This information enables smart objects to recognize context and to provide active and convenient services to the customers. However in most cases, context-aware services are available only with expert systems. In this paper, we present generalized activity recognition application in the smart home based on a naive Bayesian network(BN) and fuzzy clustering. We quantize continuous sensor data with fuzzy c-means clustering to simplify and reduce BN's conditional probability table size. And we apply mutual information to learn the BN structure efficiently. We show that this system can recognize user activities about 80% accuracy in the web based virtual smart home.