• The Journal of the Acoustical Society of Korea
• /
• v.20 no.3
• /
• pp.56-62
• /
• 2001

In this paper, we adopt a missing data theory to speech recognition. It can be used in order to maintain high performance of speech recognizer when the missing data occurs. In general, hidden Markov model (HMM) is used as a stochastic classifier for speech recognition task. Acoustic events are represented by continuous probability density function in continuous density HMM(CDHMM). The missing data theory has an advantage that can be easily applicable to this CDHMM. A marginalization method is used for processing missing data because it has small complexity and is easy to apply to automatic speech recognition (ASR). Also, a spectral subtraction is used for detecting missing data. If the difference between the energy of speech and that of background noise is below given threshold value, we determine that missing has occurred. We propose a new method that examines the reliability of detected missing data using voicing probability. The voicing probability is used to find voiced frames. It is used to process the missing data in voiced region that has more redundant information than consonants. The experimental results showed that our method improves performance than baseline system that uses spectral subtraction method only. In 452 words isolated word recognition experiment, the proposed method using the voicing probability reduced the average word error rate by 12％ in a typical noise situation.

• Journal of Korean Society of Disaster and Security
• /
• v.6 no.2
• /
• pp.23-30
• /
• 2013

A dynamic analysis of random vibration processes is concerned with the first excursion probability based on first passage time during some specified lifetime or duration of the excitation. This study is concerned with the estimation of first-passage probability for hazard fluctuate wind velocity in the major cities reflecting the recent meteorological with largest data samples (yearly 2003-2012). The basic wind speeds were standardized homogeneously to the surface roughness category C, and to 10m above the ground surface. In this paper, the hazard fluctuate wind velocities are treated as a time-independent (stationary) random process and Gaussian random processes. The first excursion probability were calculated from Poisson model based on the independent event of level crossing & two-state Markov model based on the envelopes of level crossing.

• Journal of Environmental Science International
• /
• v.19 no.1
• /
• pp.115-124
• /
• 2010

Urban watershed can be found in the visible changes in technology, the most realistic satellite images is to use the data. Satellite image data on the indicators for progress on the nature of the change of land use is consistent and repetitive information, regular observation makes possible the detailed analysis of space-time. These remote sensing techniques and the type of course and, by using the time series history, the past, the dynamic model and the randomized prediction methodology for the conversion process if the city and river basin cooperation of the space changes effectively will be able to extrapolate. For each of the main changes in river flow, depending on the area of urbanization as determined according to reproduce the duration of the relationship between the urbanization of the area and runoff can be represented as a linear polynomial expression was, if a linear expression in the two fast slew rate of 0.858 to 0.861 showed up, and fast slew rate of 0.934 to 0.974 for the polynomial are reported. Change of land use changes in the watershed of the flow is one of the most affecting elements. Therefore, changes in land use of the correct classification of rivers is a more accurate calculation of the amount of the floodgate. In particular, using the Landsat images through the image of the land use category, land use past data and calculated using the Markov Chain model and predict the future land use plan in the water control project will be used for large likely.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.38C no.2
• /
• pp.213-218
• /
• 2013

One of the most popular approaches to parameter adaptation in hidden Markov model (HMM) based systems is the maximum likelihood linear regression (MLLR) technique. In our previous study, we proposed factored MLLR (FMLLR) where each MLLR parameter is defined as a function of a control vector. We presented a method to train the FMLLR parameters based on a general framework of the expectation-maximization (EM) algorithm. Using the proposed algorithm, supplementary information which cannot be included in the models is effectively reflected in the adaptation process. In this paper, we apply the FMLLR algorithm to a pitch sequence as well as spectrum parameters. In a series of experiments on artificial generation of expressive speech, we evaluate the performance of the FMLLR technique and also compare with other approaches to parameter adaptation in HMM-based speech synthesis.

• Korean Journal of Remote Sensing
• /
• v.17 no.2
• /
• pp.183-188
• /
• 2001

The NOAA AVHRR remotely sensed SST data, collected by the National Fisheries Research and Development Institute (NFRDI), are analyzed in order to understand the spatial and temporal distributions of SST in the sea near korea. Our study is based on 10-day SST images during last 7 years (1991-1997). For a time series analysis of multiple SST images, all of images must be consistent exactly at the same position by adjusting the scales and positions of each SST image. We devised an algorithm which automatically detects cloud pixels from multiple SST images. The cloud detection algorithm is based on a physical constraint that SST anomalies in the ocean do not exceed certain limits (we used $\pm$3$^{\circ}C$ as a criterion of SST anomalies). The remotely sensed SST data are tuned by comparing remotely sensed data with observed SST at coastal stations. Seasonal variations of SST are studied by harmonic fit of SST normals at each pixel and the SST anomalies are studied by statistical method. It was found that the SST anomalies are rather persistent for one or two months. Utilizing the persistency of SST anomalies, we devised an algorithm for a prediction of future SST. In the Markov lprocess model of SST anomalies, autoregression coefficients of SST anomalies during a time elapse of 10 days are between 0.5 and 0.7. The developed algorithm with automatic cloud pixel detection and rediction of future SST is expected to be incorporated to the operational real time service of SST around Korea.

• Journal of the Korean Society for Railway
• /
• v.20 no.3
• /
• pp.349-355
• /
• 2017

Failures of railway systems can result in train delays or accidents, and therefore high reliability is required to ensure safety of railway systems. To improve reliability, railway systems are designed with redundant systems so that the standby system will continue to function normally even if the primary system fails. Generally, overall system reliability can be evaluated by the reliabilities of the parts of the whole system and the reliability of the redundant system considering common failures in case of each system is not conform physical, functional and process independent. In this study, the reliability of the hot-standby sparing system is analyzed the independent systems and dependent systems with common failures. The reliability for the standby system can be effectively analysed using Markov models, which can model the redundant configuration and the state transition.

• Journal of Creative Information Culture
• /
• v.6 no.3
• /
• pp.129-139
• /
• 2020

As the variety of scientific technology has grown repeatedly since the 19th century, innovative technology is developing high-level in healthcare field. The system to improve patient's satisfaction for silver generation introduced in domestic medical process result from promotion of convergence technology. But utilization of small and medium hospital is inevitable with maintain limited performance around existing large hospitals and high cost service system. Therefore phenomenon that weakness of patient's satisfaction and service accessibility for silver generation occurs. This study propose the design method that Android-based low-cost smart medical treatment information service system to improve accessibility to user of small and medium hospital for effective patient's satisfaction services management and medical services.

• Journal of Preventive Medicine and Public Health
• /
• v.55 no.5
• /
• pp.415-423
• /
• 2022

Economic evaluations in the healthcare are used to assess economic efficiency of pharmaceuticals and medical interventions such as diagnoses and medical procedures. This study introduces the main concepts of economic evaluation across its key steps: planning, outcome and cost calculation, modeling, cost-effectiveness results, uncertainty analysis, and decision-making. When planning an economic evaluation, we determine the study population, intervention, comparators, perspectives, time horizon, discount rates, and type of economic evaluation. In healthcare economic evaluations, outcomes include changes in mortality, the survival rate, life years, and quality-adjusted life years, while costs include medical, non-medical, and productivity costs. Model-based economic evaluations, including decision tree and Markov models, are mainly used to calculate the total costs and total effects. In cost-effectiveness or costutility analyses, cost-effectiveness is evaluated using the incremental cost-effectiveness ratio, which is the additional cost per one additional unit of effectiveness gained by an intervention compared with a comparator. All outcomes have uncertainties owing to limited evidence, diverse methodologies, and unexplained variation. Thus, researchers should review these uncertainties and confirm their robustness. We hope to contribute to the establishment and dissemination of economic evaluation methodologies that reflect Korean clinical and research environment and ultimately improve the rationality of healthcare policies.

• KIISE Transactions on Computing Practices
• /
• v.21 no.2
• /
• pp.154-159
• /
• 2015

Over the past few years, user needs in the smartphone application market have been shifted from diversity toward intelligence. Here, we propose a novel cognitive agent that plans the daily routines of users using the lifelog data collected by the smart phones of individuals. The proposed method first employs DPGMM (Dirichlet Process Gaussian Mixture Model) to automatically extract the users' POI (Point of Interest) from the lifelog data. After extraction, the POI and other meaningful features such as GPS, the user's activity label extracted from the log data is then used to learn the patterns of the user's daily routine by POMDP (Partially Observable Markov Decision Process). To determine the significant patterns within the user's time dependent patterns, collaboration was made with the SNS application Foursquare to record the locations visited by the user and the activities that the user had performed. The method was evaluated by predicting the daily routine of seven users with 3300 feedback data. Experimental results showed that daily routine scheduling can be established after seven days of lifelogged data and feedback data have been collected, demonstrating the potential of the new method of place-time-activity coupled daily routine planning systems in the intelligence application market.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.26 no.6
• /
• pp.471-476
• /
• 2016

In this paper, recognition system for continuous human action is explained by using motion history image and histogram of oriented gradient with spotter model based on depth information, and the spotter model which performs action spotting is proposed to improve recognition performance in the recognition system. The steps of this system are composed of pre-processing, human action and spotter modeling and continuous human action recognition. In pre-processing process, Depth-MHI-HOG is used to extract space-time template-based features after image segmentation, and human action and spotter modeling generates sequence by using the extracted feature. Human action models which are appropriate for each of defined action and a proposed spotter model are created by using these generated sequences and the hidden markov model. Continuous human action recognition performs action spotting to segment meaningful action and meaningless action by the spotter model in continuous action sequence, and continuously recognizes human action comparing probability values of model for meaningful action sequence. Experimental results demonstrate that the proposed model efficiently improves recognition performance in continuous action recognition system.