• Journal of Computing Science and Engineering
• /
• v.7 no.2
• /
• pp.99-111
• /
• 2013

Approximating a divergence between two probability distributions from their samples is a fundamental challenge in statistics, information theory, and machine learning. A divergence approximator can be used for various purposes, such as two-sample homogeneity testing, change-point detection, and class-balance estimation. Furthermore, an approximator of a divergence between the joint distribution and the product of marginals can be used for independence testing, which has a wide range of applications, including feature selection and extraction, clustering, object matching, independent component analysis, and causal direction estimation. In this paper, we review recent advances in divergence approximation. Our emphasis is that directly approximating the divergence without estimating probability distributions is more sensible than a naive two-step approach of first estimating probability distributions and then approximating the divergence. Furthermore, despite the overwhelming popularity of the Kullback-Leibler divergence as a divergence measure, we argue that alternatives such as the Pearson divergence, the relative Pearson divergence, and the $L^2$-distance are more useful in practice because of their computationally efficient approximability, high numerical stability, and superior robustness against outliers.

• Journal of KIISE:Computer Systems and Theory
• /
• v.31 no.3_4
• /
• pp.170-186
• /
• 2004

Recently, for real-time applications such as multimedia and signal processing, it becomes increasingly important to provide a probabilistic guarantee that each task in the application meets its deadline with a given probability. To provide the probabilistic guarantee, an analysis method is needed that can accurately predict the deadline miss probability for each task in a given system. This paper proposes a stochastic analysis method for real-time systems that use priority-driven scheduling, such as Rate Monotonic and Earliest Deadline First, in order to accurately compute the deadline miss probability of each task in the system. The proposed method accurately computes the response time distributions for tasks with arbitrary execution time distributions, and thus makes it possible to determine the deadline miss probability of individual tasks. In the paper. through experiments, we show that the proposed method is highly accurate and outperforms exisiting methods proposed in the literature.

• Transactions of the Korean Society of Mechanical Engineers A
• /
• v.32 no.9
• /
• pp.740-753
• /
• 2008

The conventional Split Hopkinson Pressure Bar (C-SHPB) technique with aluminum pressure bars to achieve a closer impedance match between the pressure bars and the specimen materials such as hot temperature degraded POM (Poly Oxy Methylene) and PP (Poly Propylene) to obtain more distinguishable experimental signals is used to obtain a dynamic behavior of material deformation under a high strain rate loading condition. An experimental modification with Pulse shaper is introduced to reduce the nonequilibrium on the dynamic material response during a short test period to increase the rise time of the incident pulse for two polymeric materials. For the dynamic stress strain curve obtained from SHPB experiment under high strain rate, the Johnson-Cook model is applied as a constitutive equation, and we verify the applicability of this constitutive equation to the probabilistic reliability estimation method. The methodology to estimate the reliability using the probabilistic method such as the FORM and the SORM has been proposed, after compose the limit state function using Johnson-Cook model. It is found that the failure probability estimated by using the SORM is more reliable than those of the FORM, and the failure probability increases with the increase of applied stress. Moreover, it is noted that the parameters of Johnson-Cook model such as A and n, and applied stress affect the failure probability more than the other random variables according to the sensitivity analysis.

• Journal of Korean Society of Steel Construction
• /
• v.15 no.5 s.66
• /
• pp.475-487
• /
• 2003

This study developed an analysis model of estimating fatigue damage using the linear elastic fracture mechanics method. Stress history occurring to an element when a truck passed over a bridge was defined as block loading and crack closure theory explaining load interaction effect was applied. Stress range frequency analysis considering dead load stress and crack opening was done. Probability of stress range frequency distribution was applied and the probability distribution parameters were estimated. The Monte Carlo simulation of generating the probability various of distribution was performed. The probability distribution of failure block numbers was obtained. With this the fatigue reliability of an element not occurring in failure could be calculated. The failure block number divided by average daily truck traffic remains the life of a day. Fatigue reliability analysis model was carried out for the welding member of cross beam flange and vertical stiffener of steel box bridge using the proposed model. Consequently, a 3.8% difference was observed between the remaining life in the peak analysis method and in the proposed analysis model. The proposed analysis model considered crack closure phase and crack retard.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.30 no.4A
• /
• pp.259-270
• /
• 2005

This paper proposes a mathematical model to analyze call-and packet-level performance of the TD-CDMA/TDD system which could serve a flexible radio resource management against multi-type heterogeneous and asymmetrical traffic conditions. On call-level analysis, the mathematical model based on queueing theory performs multi-dimensional operations using random vectors or matrices to consider multiple types of traffic and also deal with asymmetrical up- and down-direction transmissions separately. Employing the mathematical model, we obtain rail blocking probability for each type of traffic and also the optimum switching-point with the smallest call flocking probability. And on packet-level analysis, employing a non-prioritized queueing scheme between circuit and packet calls, we solve 2-dimensional random vector problem composed of the queue length for packets and the number of circuit calls being served. Finally, packet-level performance is analyzed in terms of the packet loss probability and the buffer size required under mixed-traffic conditions of multiple types of circuit and packet calls.

• Journal of the Korean Data and Information Science Society
• /
• v.28 no.4
• /
• pp.783-796
• /
• 2017

In the theory of probability, a Bernoulli trial is a random experiment with exactly two possible outcomes, "success" and "failure", in which the probability of success is the same every time the experiment is conducted. In the successive games of scissors paper stone there exists the case of draw in each game. In this paper we are interested in the ultimate success probability of each participant and the expected number of the game till any one of the two has the ultimate victory. Using our results, we can calculate the ultimate winning probability of each player of the two players and the expected number of the game till any one of the two has the ultimate victory in any case whether there is draw or not in each game.

• Structural Engineering and Mechanics
• /
• v.18 no.3
• /
• pp.343-362
• /
• 2004

Rocking response of rigid bodies with rectangular footprint, freely standing on horizontal rigid plane is studied analytically. Bodies are subjected to simulated single component of horizontal earthquakes. The effect of baseline correction, applied to simulated excitations, on the rocking response is first examined. The sensitiveness of rocking motion to the details of earthquakes and geometric properties of rigid bodies is investigated. Due to the demonstrated sensitivity of rocking response to these factors, prediction of rocking stability must be made in the framework of probability theory. Therefore, using a large number of simulated earthquakes, the effects of duration and shape of intensity function of simulated earthquakes on overturning probability of rigid bodies are studied. In the case when a rigid body is placed on any floor of a building, the corresponding probability is compared to that of a body placed on the ground. For this purpose, several shear frames are employed. Finally, the viability of the energy balance equation, which was introduced by Housner in 1963 and widely used by nuclear power industry to estimate the rocking stability of bodies, is evaluated. It is found that the equation is robust. Examples are also given to show how this equation can be used.

• Journal of KIISE:Computer Systems and Theory
• /
• v.29 no.8
• /
• pp.484-492
• /
• 2002

We propose an efficient stream authentication scheme using tree authentication. To reduce the computation cost of the sender, we design the authentication tree whose height is very short. We appropriately distribute authentication information over packets so the receiver can verify data with high probability. Moreover, we provide mathematical analysis on the verification probability. For the proposed scheme and previous schemes, we measured the elapsed time for generating authentication information and the proposed scheme has equal to or slightly larger than that of GM's scheme, which has the lowest computation overhead. We performed simulations, which show that the verification probability of the proposed scheme is much higher than that of any other scheme.

• Journal of the Korea Furniture Society
• /
• v.20 no.4
• /
• pp.339-357
• /
• 2009

Probability based design(PBD) method is independent of construction materials and uses real material properties unlike allowable stress design(ASD) that depends on small clear specimen property, also give quantitative safety and endurance lifetime of a certain material. Moreover, almost advanced country accepted PBD method instead of ASD method. So it is urgent to convert the current ASD method into the PBD method. However, there are wholly lacking of domestic researches related to current issue, and to solve several points in ASD method and to take advantage of PBD method, the conversion from the ASD method into the PBD method is a worldwide trend. Other domestic construction codes, such as steel or concrete constructions, accept the PBD method as well. Accordingly, to introduce PBD method into wood structural design, general theory, and preliminary data and methods were reviewed. With keeping this in mind, some important contents were reviewed, sorted some points for wood structural design that have distinctions against the other construction materials. Furthermore, the history of PBD method, and statistical data and theories for the PBD method, and preliminary data of resistance and load that are two random variables for the PBD method, and finally the difference between limit state design(LSD) and load and resistance factor design(LRFD) that were two superpowers in the PBD method.

• Wind and Structures
• /
• v.26 no.4
• /
• pp.231-251
• /
• 2018

The objective of the investigation is the analysis of wind-tunnel experimental errors, associated with the measurement of aeroelastic coefficients of bridge decks (Scanlan flutter derivatives). A two-degree-of-freedom experimental apparatus is used for the measurement of flutter derivatives. A section model of a closed-box bridge deck is considered in this investigation. Identification is based on free-vibration aeroelastic tests and the Iterative Least Squares method. Experimental error investigation is carried out by repeating the measurements and acquisitions thirty times for each wind tunnel speed and configuration of the model. This operational procedure is proposed for analyzing the experimental variability of flutter derivatives. Several statistical quantities are examined; these quantities include the standard deviation and the empirical probability density function of the flutter derivatives at each wind speed. Moreover, the critical flutter speed of the setup is evaluated according to standard flutter theory by accounting for experimental variability. Since the probability distribution of flutter derivatives and critical flutter speed does not seem to obey a standard theoretical model, polynomial chaos expansion is proposed and used to represent the experimental variability.