• Speech Sciences
• /
• v.10 no.2
• /
• pp.15-25
• /
• 2003

In this paper, we propose an algorithm on OOV(Out-of-Vocabulary) rejection based on two-stage method. In the first stage, the algorithm rejects OOVs using generic word model, and then in the second stage, for further reduction of false acceptance, it rejects words which have low similarity to the candidate by measuring the distance between HMM models. For the experiment, we choose 20 in-vocabulary words out of PBW445 DB distributed by ETRI. In case that the first stage is processed only, the false acceptance is 3% with 100% correct acceptance, and in case both stages are processed, the false acceptance is reduced to 1% with 100% correct acceptance.

• International Journal of Internet, Broadcasting and Communication
• /
• v.12 no.2
• /
• pp.45-50
• /
• 2020

A Language model with neural networks commonly trained with likelihood loss. Such that the model can learn the sequence of human text. State-of-the-art results achieved in various language generation tasks, e.g., text summarization, dialogue response generation, and text generation, by utilizing the language model's next token output probabilities. Monotonous and boring outputs are a well-known problem of this model, yet only a few solutions proposed to address this problem. Several decoding techniques proposed to suppress repetitive tokens. Unlikelihood training approached this problem by penalizing candidate tokens probabilities if the tokens already seen in previous steps. While the method successfully showed a less repetitive generated token, the method has a large memory consumption because of the training need a big vocabulary size. We effectively reduced memory footprint by encoding words as sequences of subword units. Finally, we report competitive results with token level unlikelihood training in several automatic evaluations compared to the previous work.

• Communications for Statistical Applications and Methods
• /
• v.21 no.6
• /
• pp.501-512
• /
• 2014

Stylized facts on asset return are fat-tail, asymmetry, volatility clustering and structure changes. This paper simultaneously captures these characteristics by introducing a multi-regime models: Finite mixture distribution and regime switching GARCH model. Analyzing the daily KOSPI return from $4^{th}$ January 2000 to $30^{th}$ June 2014, we find that a two-component mixture of t distribution is a good candidate to describe the shape of the KOSPI return from unconditional and conditional perspectives. Empirical results suggest that the equality assumption on the shape parameter of t distribution yields better discrimination of heterogeneity component in return data. We report the strong regime-dependent characteristics in volatility dynamics with high persistence and asymmetry by employing a regime switching GJR-GARCH model with t innovation model. Compared to two sub-samples, Pre-Crisis (January 2003 ~ December 2007) and Post-Crisis (January 2010 ~ June 2014), we find that the degree of persistence in the Pre-Crisis is higher than in the Post-Crisis along with a strong asymmetry in the low-volatility (high-volatility) regime during the Pre-Crisis (Post-Crisis).

• Corrosion Science and Technology
• /
• v.5 no.6
• /
• pp.201-205
• /
• 2006

HYPER(Hybrid Power Extraction Reactor) is the accelerator driven subcritical transmutation system developed by KAERI(Korea Atomic Research Institute). HYPER is designed to transmute long-lived transuranic actinides and fission products such as Tc-99 and I-129. Liquid lead-bismuth eutectic (LBE). Has been a primary candidate for coolant and spallation neutron target due to its appropriate thermal-physical and chemical properties, However, it is very corrosive to the common steels used in nuclear installations at high temperature. This corrosion problem is one of the main factors considered to set the upper limits of temperature and velocity of HYPER system. In this study, a parametric study for a corrosion model was performed. And a preliminary corrosion model was also developed to predict the corrosion rate in isothermal Pb and LBE flow loops.

• Journal of the Korean Society of Systems Engineering
• /
• v.14 no.2
• /
• pp.49-57
• /
• 2018

The main objective of this paper is come up with methodology approach for FPGA-based system in verification and validation lifecycle regarding software reliability using system engineering approach. The steps of both reverse engineering and re-engineering are carried out to implement an FPGA-based of safety critical system in Nuclear Power Plant. The reverse engineering methodology is applied to elicit the requirements of the system as well as gain understanding of the current life cycle and V&V activities of FPGA based-system. The re-engineering method is carried out to get a new methodology approach of software reliability, particularly Software Reliability Growth Model. For measure the software reliability of a given FPGA-based system, the following steps are executed as; requirements definition and measurement, evaluation of candidate reliability model, and the validation of the selected system. As conclusion, a new methodology approach for software reliability measurement using software reliability growth model is developed.

• Smart Media Journal
• /
• v.4 no.4
• /
• pp.18-23
• /
• 2015

In this paper, we propose a method for detecting the number of clusters. This method can improve the performance of a gaussian mixture model function in conventional markov random field method by using the tensor voting. The key point of the proposed method is that extracts the number of the center through the continuity of saliency map of the input data of the tensor voting token. At first, we separate the foreground and background region candidate in a given natural images. After that, we extract the appropriate cluster number for each separate candidate regions by applying the tensor voting. We can make accurate modeling a gaussian mixture model by using a detected number of cluster. We can return the result of natural binary text image by calculating the unary term and the pairwise term of markov random field. After the experiment, we can confirm that the proposed method returns the optimal cluster number and text binarization results are improved.

• Korean Journal of Construction Engineering and Management
• /
• v.10 no.5
• /
• pp.148-157
• /
• 2009

When an emergency situation happens in buildings, the top priority is to ensure the occupant from danger as soon as possible. Achieving that goal is a multifaceted and difficult task. However, current evacuation systems have many deficiencies in dealing with the emergency in multi-level structures. The shortage of abilities to continuously update database, predict the future situation and provide the information to users with contextual information is the limit in current systems. Thus, it is very crucial to introduce Evacuation Information System (EIS), which is able to respond quickly to the emergency, and transfer the information to both the administrator and the occupant. The main purpose of this paper is to build EIS on the basis of the indoor Geographical Information System (GIS). When the emergency happens, EIS gives the instruction to Emergency Response Model (ERM) at once. ERM carries out the order and calculates the optimal evacuation routes, then sends the result to EIS. At last, EIS transmits evacuation messages to the occupant who implements evacuation plan. This paper highlights the benefits of EIS in two aspects. One is that EIS can update the data continuously to support evacuation strategy-making. The other is that it can transmit evacuation messages to both the administrator and the occupant.

• Tunnel and Underground Space
• /
• v.18 no.2
• /
• pp.134-148
• /
• 2008

The state of in situ stress is an important factor in considering the suitability of a site as a geological repository for nuclear waste. In this study, three-dimensional distinct numerical analysis was conducted to investigate the effect of deformation zones on the state of stress in the Oskarshamn area, which is one of two candidate sites in Sweden. A discontinuum numerical model was constructed by explicitly representing the numerous deformation zones identified from site investigation and far-field tectonic stress was applied in the constructed model. The numerical model successfully captured the variation of measured stress often observed in the rock mass containing large-scale fractures, which shows that numerical analysis can be an effective tool in improving the understanding of the state of stresses. Discrepancies between measured and modelled stress are attributed to the inconsistent quality of measured stress, uncertainty in geological geometry. and input data for fractures.

• Nuclear Engineering and Technology
• /
• v.40 no.5
• /
• pp.387-396
• /
• 2008

Many experimental analyses for annular film dryouts, which is one of the Critical Heat Flux (CHF) mechanisms, have been performed because of their importance. Numerical approaches must also be developed in order to assess the results from experiments and to perform pre-tests before experiments. Various thermal-hydraulic codes, such as RELAP, COBRATF, MARS, etc., have been used in the assessment of the results of dryout experiments and in experimental pre-tests. These thermal-hydraulic codes are general tools intended for the analysis of various phenomena that could appear in nuclear power plants, and many models applying these codes are unnecessarily complex for the focused analysis of dryout phenomena alone. In this study, a numerical model was developed for annular film dryout using the drift-flux model from uniform heated tube geometry. Several candidates of models that strongly affect dryout, such as the entrainment model, deposition model, and the criterion for the dryout point model, were tested as candidates for inclusion in an optimized annular film dryout model. The optimized model was developed by adopting the best combination of these candidate models, as determined through comparison with experimental data. This optimized model showed reasonable results, which were better than those of MARS code.

• Korean Journal of Veterinary Service
• /
• v.43 no.2
• /
• pp.89-97
• /
• 2020

This study was carried out to examine a novel inactivated Salmonella Typhimurium (S. Typhimurium) vaccine candidate for protection of mice against salmonellosis by immunization of BALB/c mice using various type adjuvant. The novel type-inactivated vaccine candidate was constructed by adding Chlorhexidine digluconate solution. BALB/c mice were divided into 6 groups of 15 mice apiece. The mice were intramuscularly (IM) primed at 6 weeks of age and were IM boosted 8 weeks of age. Groups A and B mice were injected with sterile phosphate-buffered saline as controls; group C mice were inoculated with 5×10⁸ cells/100 µL of formalin-inactivated S. Typhimurium cells and adjuvant ISA70; groups D~F mice were immunized with 5×10⁸ cells/100 µL of the inactivated vaccine candidate and adjuvant ISA70, adjuvant IMS1313 and adjuvant IMS1313 containing 30 ㎍/mL of GI24, respectively. All mice (except group A mice) were orally challenged with a virulent S. Typhimurium strain at 10 weeks of age. Mice from groups C-F had significantly increased IgG levels compared to control groups (A-B) mice. The levels of splenocyte IFN-γ and IL-4 in mice of all groups were measured by ELISA, resulting in increased immunity in group F mice compared to those of groups A-E mice. These data suggested that systemic and cell-mediated immune responses were highly induced by IM immunization with the vaccine candidate and adjuvant IMS1313 containing GI24. Furthermore, clinical signs such as death were observed in only 20% of group F mice after virulent Salmonella strain challenge, however, groups B and C (100%), and groups D and E (60%) mice died. This data suggested that mice immunized by intramuscular prime and booster with this vaccine candidate and adjuvant IMS1313 containing GI24 effectively protected mice from salmonellosis.