• The KIPS Transactions:PartB
• /
• v.13B no.4 s.107
• /
• pp.383-392
• /
• 2006

In this paper, we propose a new relevance feedback algorithm using Probabilistic Neural Networks(PNN) while supporting multi-class learning. Then, to validate the effectiveness of our feedback approach, we incorporate the proposed algorithm into our region-based image retrieval tool, FRIP(Finding Regions In the Pictures). In our feedback approach, there is no need to assume that feature vectors are independent, and as well as it allows the system to insert additional classes for detail classification. In addition, it does not have a long computation time for training because it only has four layers. In the PNN classification process, we store the user's entire past feedback actions as a history in order to improve performance for future iterations. By using a history, our approach can capture the user's subjective intension more precisely and prevent retrieval performance errors which originate from fluctuating or degrading in the next iteration. The efficacy of our method is validated using a set of 3000 images derived from a Corel-photo CD.

• The KIPS Transactions:PartB
• /
• v.9B no.6
• /
• pp.791-798
• /
• 2002

In this thesis. we present a new method for inducing a probabilistic dependency grammar (PDG) from text corpus. As words in Korean are composed of a set of more basic morphemes, there exist various dependency relations in a word. So, if the induction process does not take into account of these in-word dependency relations, the accuracy of the resulting grammar nay be poor. In comparison with previous PDG induction methods. the main difference of the proposed method lies in the fact that the method takes into account in-word dependency relations as well as inter-word dependency relations. To access the performance of the proposed method, we conducted an experiment using a manually-tagged corpus of 25,000 sentences which is complied by Korean Advanced Institute of Science and Technology (KAIST). The grammar induction produced 2,349 dependency rules. The parser with these dependency rules shoved 69.77% accuracy in terms of the number of correct dependency relations relative to the total number dependency relations for best-1 parse trees of sample sentences. The result shows that taking into account in-word dependency relations in the course of grammar induction results in a more accurate dependency grammar.

• Journal of the Korean Society of Safety
• /
• v.32 no.3
• /
• pp.160-171
• /
• 2017

US national research laboratories developed the first Vital Area Identification (VAI) method for the physical protection of nuclear power plants that is based on Event Tree Analysis (ETA) and Fault Tree Analysis (FTA) techniques in 1970s. Then, Korea Atomic Energy Research Institute proposed advanced VAI method that takes advantage of fire and flooding Probabilistic Safety Assessment (PSA) results. In this study, in order to minimize the burden and difficulty of VAI, (1) a set of streamlined VAI rules were developed, and (2) this set of rules was applied to PSA fault tree and event tree at the initial stage of VAI process. This new rule-based VAI method is explained, and its efficiency and correctness are demonstrated throughout this paper. This new rule-based VAI method drastically reduces problem size by (1) performing PSA event tree simplification by applying VAI rules to the PSA event tree, (2) calculating preliminary prevention sets with event tree headings, (3) converting the shortest preliminary prevention set into a sabotage fault tree, and (4) performing usual VAI procedure. Since this new rule-based VAI method drastically reduces VAI problem size, it provides very quick and economical VAI procedure. In spite of an extremely reduced sabotage fault tree, this method generates identical vital areas to those by traditional VAI method. It is strongly recommended that this new rule-based VAI method be applied to the physical protection of nuclear power plants and other complex safety-critical systems such as chemical and military systems.

• The Journal of the Acoustical Society of Korea
• /
• v.39 no.2
• /
• pp.137-142
• /
• 2020

Many studies based on I-vector have been conducted in a variety of environments, from text-dependent short-utterance to text-independent long-utterance. In this paper, we propose a speaker verification system employing a combination of I-vector with Probabilistic Linear Discriminant Analysis (PLDA) and speaker embedding of Long Short Term Memory (LSTM) with attention mechanism in far-field and noisy environments. The LSTM model's Equal Error Rate (EER) is 15.52 % and the Attention-LSTM model is 8.46 %, improving by 7.06 %. We show that the proposed method solves the problem of the existing extraction process which defines embedding as a heuristic. The EER of the I-vector/PLDA without combining is 6.18 % that shows the best performance. And combined with attention-LSTM based embedding is 2.57 % that is 3.61 % less than the baseline system, and which improves performance by 58.41 %.

• Journal of the Computational Structural Engineering Institute of Korea
• /
• v.16 no.3
• /
• pp.281-290
• /
• 2003

Co Evolutionary Structural Design(CESD) Framework is presented, which can deal with the load design and structural topology design simultaneously. The load design here is the exploration algorithm that finds the critical load patterns of the given structure. In general, the load pattern is a crucial factor in determining the structural topology and being selected from the experts어 intuition and experience. However, if any of the critical load patterns would be excluded during the process of problem formation, the solution structure might show inadequate performance under the load pattern. Otherwise if some reinforcement method such as safety factor method would be utilized, the solution structure could result in inefficient conservativeness. On the other hand, the CESD has the ability of automatically finding the most critical load patterns and can help the structural solution evolve into the robust design. The CESD is made up of a load design discipline and a structural topology design discipline both of which have the fully coupled relation each other. This coupling is resolved iteratively until the resultant solution can resist against all the possible load patterns and both disciplines evolve into the solution structure with the mutual help or competition. To verify the usefulness of this approach, the 10 bar truss and the jacket type offshore structure are presented. SORA(Sequential Optimization & Reliability Assessment) is adopted in CESD as a probabilistic optimization methodology, and its usefulness in decreasing the computational cost is verified also.

• Journal of the Korea Society of Computer and Information
• /
• v.11 no.6 s.44
• /
• pp.261-267
• /
• 2006

In order to perform a probabilistic safety assessment (PSA), it requires a large number of data for various fields. And the quality of a PSA results have become more important thing of the risk assessment. As part of enhancing the PSA qualify, Korea Atomic Energy Research Institute is developing a full power Human Reliability Analysis (HRA) calculator to manage human failure events (HFEs) and to calculate the diagnosis human error probabilities and execution human error probabilities. This paper introduces the development process and an overview of a standard HRA method for nuclear power plants. The study was carried out in three stages; 1) development of the procedures and rules for a standard HRA method. 2) design of a system structure, 3) development of the HRA calculator.

• Journal of the Korean Solar Energy Society
• /
• v.34 no.4
• /
• pp.23-29
• /
• 2014

Recently, in power system studies, Multiple Energy Carriers (MECs) such as Energy Hub has been broadly utilized in power system planners and operators. Particularly, Energy Hub performs one of the most important role as the intermediate in implementing the MECs. However, it still needs to be put under examination in both modeling and operating concerns. For instance, a probabilistic optimization model is treated by a robust global optimization technique such as multi-agent genetic algorithm (MAGA) which can support the online economic dispatch of MECs. MAGA also reduces the inevitable uncertainty caused by the integration of selected input energy carriers. However, MAGA only considers current state of the integration of selected input energy carriers in conjunctive with the condition of smart grid environments for decision making in Energy Hub. Thus, in this paper, we propose an immune algorithm based Multiple Energy Carriers System which can adopt the learning process in order to make a self decision making in Energy Hub. In particular, the proposed immune algorithm considers the previous state, the current state, and the future state of the selected input energy carriers in order to predict the next decision making of Energy Hub based on the probabilistic optimization model. The below figure shows the proposed immune algorithm based Multiple Energy Carriers System. Finally, we will compare the online economic dispatch of MECs of two algorithms such as MAGA and immune algorithm based MECs by using Real Time Digital Simulator (RTDS).

• Journal of the Earthquake Engineering Society of Korea
• /
• v.7 no.1
• /
• pp.31-40
• /
• 2003

Due to the random nature of earthquake, the definition of the input excitation is one of the major uncertainties in the seismic response analysis. Furthermore, ground motions that correspond to a limited number of design parameters are not unique. Consequently, a brood range of response values can be obtained even with a set of motions, which match the same target parameters. The paper presents a practical probabilistic approach that can be used to systematically model the stochastic nature of seismic loading. The new approach is based on energy-based RMS hazard and takes account for the uncertainties of key ground motion parameters. The simulations indicate that the new RMS procedure is particularly useful for the rigorous probabilistic seismic response analysis, since the procedure is suitable for generation of large number of hazard-compatible motions, unlike the conventional procedure that aim to generate a small number of motions.

• Journal of Korea Water Resources Association
• /
• v.41 no.6
• /
• pp.629-641
• /
• 2008

Many studies on sustained droughts have often been limited to the analysis of historic flow series. A major disadvantage in this approach can be described as the lack of long historic flow records needed to obtain a significant number of drought events for the analysis. To overcome this difficulty, one of the present study idea is to use synthetically generated hydrologic series. A methodology is presented to develop flow series based on the probabilistic analysis of the stochastic properties of the observed flows. The method can be utilized to generate a flow series of desired length so as to include many multiyear drought events within the process. In this paper, a concept of creating multiyear drought scenarios is introduced, and its development procedure is illustrated by a case study of the water supply system in Han River Basin. Also, it was found that the generated flow series can be reliably used to predict the long drought duration and sustained drought hydrologic scenarios within a given return period.

• International Journal of Highway Engineering
• /
• v.10 no.3
• /
• pp.199-208
• /
• 2008

Prediction the performance of pavement provides proper information to an agency on decision-making process; especially evaluating the pavement performance and prioritizing the work plan. To date, there are a number of approaches to predict the future deterioration of pavements. However, there are some limitation to proper prediction of the pavement service life. In this paper, pavement performance model and pavement condition prediction model are developed in order to improve pavement condition prediction method. The prediction model of pavement condition through the regression analysis of real pavement condition is based on the probability distribution of pavement condition, which set to 5%, 15%, 25% and 50%, by condition of the pavement and traffic volume. The pavement prediction model presented from the behavior of individual pavement condition which are set to 5%, 15%, 25% and 50% of probability distribution. The performance of the prediction model is evaluated from analyzing the average, standard deviation of HPCI, and the percentage of HPCI which is lower than 3.0 of comparable section. In this paper, we will suggest the more rational method to determine the future pavement conditions, including the probabilistic duration and deterministic modeling methods regarding the impact of traffic volume, age, and the type of the pavement.