• Journal of Wetlands Research
• /
• v.11 no.1
• /
• pp.49-64
• /
• 2009

Rainfall-runoff models are used for efficient management, distribution, planning, and design of water resources in accordance with the process of hydrologic cycle. The models simplify the transition of rainfall to runoff as rainfall through different processes including evaporation, transpiration, interception, and infiltration. As the models simplify complex physical processes, gaps between the models and actual rainfall events exist. For more accurate simulation, appropriate models that suit analysis goals are selected and reliable long-term hydrological data are collected. However, uncertainty is inherent in models. It is therefore necessary to evaluate reliability of simulation results from models. A number of studies have evaluated uncertainty ingrained in rainfall-runoff models. In this paper, Meta-Gaussian method proposed by Montanari and Brath(2004) was used to assess uncertainty of simulation outputs from rainfall-runoff models. The model, which estimates upper and lower bounds of the confidence interval from probabilistic distribution of a model's error, can quantify global uncertainty of hydrological models. In this paper, Meta-Gaussian method was applied to analyze uncertainty of simulated runoff outputs from $Vflo^{TM}$, a physically-based distribution model and HEC-HMS model, a conceptual lumped model.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.7 no.12
• /
• pp.3274-3292
• /
• 2013

Unsupervised methods for image segmentation are recently drawing attention because most images do not have labels or tags. A topic model is such an unsupervised probabilistic method that captures latent aspects of data, where each latent aspect, or a topic, is associated with one homogeneous region. The results of topic models, however, usually have noises, which decreases the overall segmentation performance. In this paper, to improve the performance of image segmentation using topic models, we propose two topic masks applicable to topic assignments of homogeneous regions obtained from topic models. The topic masks capture the noises among the assigned topic assignments or topic labels, and remove the noises by replacements, just like image masks for pixels. However, as the nature of topic assignments is different from image pixels, the topic masks have properties that are different from the existing image masks for pixels. There are two contributions of this paper. First, the topic masks can be used to reduce the noises of topic assignments obtained from topic models for image segmentation tasks. Second, we test the effectiveness of the topic masks by applying them to segmented images obtained from the Latent Dirichlet Allocation model and the Spatial Latent Dirichlet Allocation model upon the MSRC image dataset. The empirical results show that one of the masks successfully reduces the topic noises.

• Journal of Radiation Protection and Research
• /
• v.42 no.2
• /
• pp.130-140
• /
• 2017

Background: With the need for a domestic level 3 probabilistic safety assessment (PSA), it is essential to develop a Korea-specific code. Health effect assessments study radiation-induced impacts; in particular, long-term health effects are evaluated in terms of cancer risk. The objective of this study was to analyze the latest cancer risk models developed by foreign organizations and to compare the methodology of how they were developed. This paper also provides suggestions regarding the development of Korean cancer risk models. Materials and Methods: A review of cancer risk models was carried out targeting the latest models: the NUREG model (1993), the BEIR VII model (2006), the UNSCEAR model (2006), the ICRP 103 model (2007), and the U.S. EPA model (2011). The methodology of how each model was developed is explained, and the cancer sites, dose and dose rate effectiveness factor (DDREF) and mathematical models are also described in the sections presenting differences among the models. Results and Discussion: The NUREG model was developed by assuming that the risk was proportional to the risk coefficient and dose, while the BEIR VII, UNSCEAR, ICRP, and U.S. EPA models were derived from epidemiological data, principally from Japanese atomic bomb survivors. The risk coefficient does not consider individual characteristics, as the values were calculated in terms of population-averaged cancer risk per unit dose. However, the models derived by epidemiological data are a function of sex, exposure age, and attained age of the exposed individual. Moreover, the methodologies can be used to apply the latest epidemiological data. Therefore, methodologies using epidemiological data should be considered first for developing a Korean cancer risk model, and the cancer sites and DDREF should also be determined based on Korea-specific studies.

• Journal of KIISE:Software and Applications
• /
• v.34 no.12
• /
• pp.1045-1055
• /
• 2007

Latent topic models are statistical models which automatically captures salient patterns or correlation among features underlying a data collection in a probabilistic way. They are gaining an increased popularity as an effective tool in the application of automatic semantic feature extraction from text corpus, multimedia data analysis including image data, and bioinformatics. Among the important issues for the effectiveness in the application of latent topic models to the massive data set is the efficient learning of the model. The paper proposes an accelerated learning technique for PLSA model, one of the popular latent topic models, by an incremental EM algorithm instead of conventional EM algorithm. The incremental EM algorithm can be characterized by the employment of a series of partial E-steps that are performed on the corresponding subsets of the entire data collection, unlike in the conventional EM algorithm where one batch E-step is done for the whole data set. By the replacement of a single batch E-M step with a series of partial E-steps and M-steps, the inference result for the previous data subset can be directly reflected to the next inference process, which can enhance the learning speed for the entire data set. The algorithm is advantageous also in that it is guaranteed to converge to a local maximum solution and can be easily implemented just with slight modification of the existing algorithm based on the conventional EM. We present the basic application of the incremental EM algorithm to the learning of PLSA and empirically evaluate the acceleration performance with several possible data partitioning methods for the practical application. The experimental results on a real-world news data set show that the proposed approach can accomplish a meaningful enhancement of the convergence rate in the learning of latent topic model. Additionally, we present an interesting result which supports a possible synergistic effect of the combination of incremental EM algorithm with parallel computing.

• 제어로봇시스템학회:학술대회논문집
• /
• 1991.10b
• /
• pp.1597-1600
• /
• 1991

In this paper, we describe a notion of sensor modeling method in multisensor data fusion using fuzzy set theory. Each sensor module is characterized by its fuzzy constraints to specific features of environment. These sensor fuzzy constraints can be imposed on multisensory data to verify their degree of truth and compatibility toward the final decision making. In comparison with other sensor modeling methods, such as probabilistic models or rule-based models, the proposed method is very simple and can be easily implemented in intelligent robot systems.

• Water for future
• /
• v.4 no.1
• /
• pp.51-58
• /
• 1971

Hydrologic data serve as an input to the water resources system. An adequate analysis of hydrologic data is one of the most important steps in the planning of the water resources development program. The natural hydrologic processes, which produce the hydrologic data, are truely 'stochastic' in the sense that natural hydrologic phenomena change with time in accordance with the law of probability as well as with sequential relationship between their occurrences. Therefore, the stochastic approach to the analysis of hydrologic data has become more popular in recent years than the conventional deterministic or probabilistic approach. This paper reviews the mathematical models which can adequately simulate the stochastic behavior of the hydrologic characteristics of a hydrologic system. The actual application of these models in the analysis of hydrologic records(precipipitation and runoff records in particular) is also presented.

• Proceedings of the Korean Operations and Management Science Society Conference
• /
• 2005.05a
• /
• pp.1114-1116
• /
• 2005

Shop floor control systems (SFCS) are used to make real-time planning and scheduling decisions to optimize the efficiency of manufacturing shops. These shops exhibit a non-linear, dynamic evolution caused by 1) the concurrent flows of disparate parts following complex routings, 2) a variety of machines that breakdown at random times, 3) stochastic arrivals of new parts with different priorities, and 4) jobs that have probabilistic processing times and transportation times. Because of their ability to capture that evolution faithfully, simulation models are often used in the aforementioned decisions. In this paper, various types of decision-making problems encountered in a shop floor have been investigated and categorized into process related problems and resource related problems for shop floor simulation.

• Journal of the Korean Statistical Society
• /
• v.29 no.4
• /
• pp.407-422
• /
• 2000

This article is concerned with the selecting predictor variables to be included in building a class of binary response t-link regression models where both probit and logistic regression models can e approximately taken as members of the class. It is based on a modification of the stochastic search variable selection method(SSVS), intended to propose and develop a Bayesian procedure that used probabilistic considerations for selecting promising subsets of predictor variables. The procedure reformulates the binary response t-link regression setup in a hierarchical truncated normal mixture model by introducing a set of hyperparameters that will be used to identify subset choices. In this setup, the most promising subset of predictors can be identified as that with highest posterior probability in the marginal posterior distribution of the hyperparameters. To highlight the merit of the procedure, an illustrative numerical example is given.

• Proceedings of the IEEK Conference
• /
• 2009.05a
• /
• pp.264-266
• /
• 2009

A skill is defined as the special ability to do something well, especially as acquired by learning and practice. To learn a skill, a Bayesian network model for representing the skill is first learned. We will regard the Bayesian network for a skill as an affordance. We propose a soft Behavior Motivation(BM) switch as a method for ordering affordances to accomplish a task. Then, a skill is constructed as a combination of an affordance and a soft BM switch. To demonstrate the validity of our proposed method, some experiments were performed with GENIBO(Pet robot) performing a task using skills of Search-a-target-object, Approach-a-target-object, Push-up-in front of -a-target-object.

• Journal of Computing Science and Engineering
• /
• v.3 no.3
• /
• pp.143-164
• /
• 2009

This article surveys recent research in the area of language modeling (sometimes called statistical language modeling) approaches to information retrieval. Language modeling is a formal probabilistic retrieval framework with roots in speech recognition and natural language processing. The underlying assumption of language modeling is that human language generation is a random process; the goal is to model that process via a generative statistical model. In this article, we discuss current research in the application of language modeling to information retrieval, the role of semantics in the language modeling framework, cluster-based language models, use of language modeling for XML retrieval and future trends.