• Journal of Korea Multimedia Society
• /
• v.23 no.7
• /
• pp.883-890
• /
• 2020

In this paper, we propose a variational expectation-maximization algorithm that computes posterior probabilities from Latent Dirichlet Allocation (LDA) model. The algorithm approximates the intractable posterior distribution of a document term matrix generated from a corpus made up by 50 papers. It approximates the posterior by searching the local optima using lower bound of the true posterior distribution. Moreover, it maximizes the lower bound of the log-likelihood of the true posterior by minimizing the relative entropy of the prior and the posterior distribution known as KL-Divergence. The experimental results indicate that documents clustered to image classification and segmentation are correlated at 0.79 while those clustered to object detection and image segmentation are highly correlated at 0.96. The proposed variational inference algorithm performs efficiently and faster than Gibbs sampling at a computational time of 0.029s.

• Proceedings of the Acoustical Society of Korea Conference
• /
• 1984.12a
• /
• pp.100-106
• /
• 1984

Traditional methods for estimating the location of underwater target, i.e. the triangulation method and the wavefront curvature method, have been utilized. The location of a target is defined by the range and the bearing, which estimates can be obtained by evaluating the time delay between neighboring sensors. Many components of error occur in estimating the target range, among which the error due to the fluctuation of heading angle is outstanding. In this paper, the wavefront curvature method was used. We considered the error due to the heading fluctuation as the $\beta$-density process, from which we analized the range estimates with $\beta$-density function exist in some finite limits, and its mean value and variation are depicted as a function of true range and heading fluctuation. Given heading angles and sensor separation, maximum estimated heading errors are presented as a function of true range.

• Proceedings of the Korean Society of Precision Engineering Conference
• /
• 1995.10a
• /
• pp.3-7
• /
• 1995

Force generated during grinding process causes elastic defomation. The effect of this deforms a workpiecs. So grinding system is explainable using the concept of macining elasticity phenomenon. Machining elasticity is defined as ratio between the true depth of c ut, and an importnat factor to affect material removal mchanism and productivity. Generally, to produce accurate surface and dimensionally precise components operators depend on their experiences. Because of these, productivity is reduced and time is wasted. The objective of this reserch is to study the effect of grinding conditions, such as table speed, depth of cut on the machining elasticity parameter.

• 한국정보디스플레이학회:학술대회논문집
• /
• 2003.07a
• /
• pp.664-667
• /
• 2003

Super In Plane Switching (S-IPS) technology is applied for large TFT-LCD panels used in TV applications. It has a lot of advantages in comparison to the alternative, VA technology. S-IPS shows excellent viewing angle properties and fast response time between intermediate gray levels. If the performance parameters which describe the actual visual performance are considered, S-IPS is much more advantageous.$^{1)}$ However, it shows relatively low contrast ratio in diagonal direction compared to viewing angle characteristic in upper/down direction in S-IPS. In order to compensate the relatively low diagonal contrast ratio, a newly designed optical film was applied and the truly wide view angle of a S-IPS TW(True Wide)-IPS) was achieved. Our newly developed 30-inch TFT-LCD panel reveals TW-IPS that is optimized for TV application

• Journal of the Korean Institute of Telematics and Electronics B
• /
• v.32B no.5
• /
• pp.777-787
• /
• 1995

This paper presents a new method for automatically calibrating robot link (Kinematic) parameters during the process of estimating motion parameters of a moving object. The motion estimation is performed based on stereo cameras mounted on the end-effector of a robot manipulator. This approach significantly differs from other calibration approaches in that the calibration is achieved by simply observing the motion of the moving object (without resorting to any other external calibrating tools) at numerous and widely varying joint-angle configurations. A differential error model, which expresses the measurement errors of a robot in terms of robot link parameter errors and motion parameters, is developed. And then a measurement equation representing the true measurement values is derived. By estimating the above two kinds of parameters minimizing the difference between the measurement equations and the true moving pattern, the calibration of the robot link parameters and the estimation of the motion parameters are accomplished at the same time.

• KOREAN JOURNAL OF CROP SCIENCE
• /
• v.35 no.6
• /
• pp.548-558
• /
• 1990

Physiology and genetics of early maturity in cereals are the subject of practical as well as scientific interest for agronomist and plant breeders, Thorough understanding of the true nature of such a complex character requires physiological and genetical knowledge about the internal factors, which are closely bound up with and react to some particular external or environmental factors. From the practical point of view. experiments should be conducted under controlled conditions. especially the day length and temperature, so that the genotypic differences pertaining to these factors may be discerned. Takahashi and Yasuda (1958, 1970) maintained that at least three physiological factors were responsible for determining earliness in barley. namely. (1) spring and winter habit of growth or vernalization requirement, (2) ogitioeruiduc response or sensitivity to short-day, and (3) earliness factor in a narrow sense or minimal vegetative growth. The same situations were true in common wheat also (Yasuda and Shimoyama, 1965), In this report. physiology and genetics of internal factors and their relations to the time of heading in the field will be presented with some problems concerning differences in mechanism of early maturity between barley and wheat.

• The Bulletin of The Korean Astronomical Society
• /
• v.43 no.1
• /
• pp.61.1-61.1
• /
• 2018

Outflows are common among local AGNs. Woo et al. (2017) suggested that AGN feedback through outflows is delayed by a dynamical time scale before the suppression of SFR is observationally detected. However, these SFR have large uncertainties because they were estimated by Artificial Neural Network (ANN) method (Ellison et al. 2016). We measured the SFR of 21 far-IR matched sources (z < 0.1) with total IR luminosity from multi-wavelength SED fitting from UV to radio. 15 out of 21 sources were observed with JCMT SCUBA-2 450 and 850um and 4 and 2 sources were matched with archival data of JCMT SCUBA-2 and Herschel SPIRE, respectively. We compared the true SFR by SED fitting with ANN-based one. In addition, we confirmed that sub-mm data are important to determine the SFR with total IR luminosity from SED fitting. Finally, we discuss the significance of true SFR and further the AGN-SF link.

• Journal of the Korean Mathematical Society
• /
• v.50 no.4
• /
• pp.879-897
• /
• 2013

Shannon entropy is one of the widely used randomness measures especially for cryptographic applications. However, the conventional entropy tests are less sensitive to the inter-bit dependency in random samples. In this paper, we propose new online randomness test schemes for true random number generators (TRNGs) based on the mutual information between consecutive ${\kappa}$-bit output blocks for testing of inter-bit dependency in random samples. By estimating the block entropies of distinct lengths at the same time, it is possible to measure the mutual information, which is closely related to the amount of the statistical dependency between two consecutive data blocks. In addition, we propose a new estimation method for entropies, which accumulates intermediate values of the number of frequencies. The proposed method can estimate entropy with less samples than Maurer-Coron type entropy test can. By numerical simulations, it is shown that the new proposed scheme can be used as a reliable online entropy estimator for TRNGs used by cryptographic modules.

• International Journal of Reliability and Applications
• /
• v.11 no.2
• /
• pp.69-87
• /
• 2010

The optimal replacement problem was investigated for a multi-state deteriorated system for which the true internal state cannot be observed directly except when the system breaks down completely. The internal state was assumed to be monitored incompletely by a monitor that gives information related to the true state of the system. The problem was formulated as a partially observable Markov decision process. The optimal procedure was found to be a monotone procedure with respect to stochastic increasing ordering of the state probability vectors under some assumptions. Limiting the optimal procedure to a monotone procedure would greatly reduce the tremendous amount of calculation time required to find the optimal procedure.

• Journal of the Korean Operations Research and Management Science Society
• /
• v.9 no.2
• /
• pp.23-27
• /
• 1984

We Observe a sequence of i. i. d random variables with density f or g. Only if g is true we should stop the process. Hence. the testing problem is completely described by a stopping time. Among all stopping times with error probability of first kind not exceeding a given bound, the one-sided sequential probability ratio test has smallest expected sample size if g is true. Moreover, the generalized one-sided SPRT has smallest expected sample size for g in the class of stopping times with expected sample size under f not falling below a given bound.