Papers by Keyword: Bayesian

Abstract: Assessment model of the vulnerability for information system is improved by using Bayesian equilibrium algorithm. The mathematical evaluation model of combined complex network information systems is established through the combination of weighted directional algorithm, and the algorithmic routine of network vulnerability assessment is designed. In order to verify the validity and reliability of the model and the algorithm, the test platform of complex network is built, and the vulnerability of network is detected with the weighted directional method, which has got the probability distribution nephogram of network vulnerability and the curve of network performance with time changing. At the last, the effect of different nodes of the network on the vulnerability of system is calculated with directed weights. And the results shown that the attacked number of different nodes and the attacked probability have improved the credibility of information analysis, which has provided theory reference for the research of information system vulnerability.

Abstract: A method for online identifying and processing single variable outliers was proposed based on a three-sliding window-Bayesian method. Generally, the method utilized the characteristic that the flow rate and temperature in metallurgical production do not change suddenly. Based on this characteristic, the research accurately identified outliers and variation of normal working points by analyzing the change of Bayesian posterior probability and conditional probability of the detection data in the three sliding windows.

Abstract: Empirical acoustic models were developed for dense-graded and open-graded asphalt concrete. Tire/pavement noise data were collected from in-service flexible pavements at different frequency bands for four consecutive years. These data were panel structured, and with a portion of observations missed arbitrarily. A Monte Carlo Markov Chain (MCMC) sampling and a multiple imputation (MI) algorithm were used to capture the unobserved heterogeneity and deal with missing observations by Bayesian simulations that are associated with the data. Models for the two mixes at different frequency bands were constructed. Major findings of the study include: first, tire/pavement noise increases with age at all frequency bands; second, tire/pavement noise level increases with air-void content of the surface mixes at medium and high frequencies but decreases at low frequencies; third, tire/pavement noise level increases with mean profile depth (MPD) at low and medium frequencies but decreases at high frequencies; and fourth, open-graded mix has low noise level compared to its dense-graded counterpart.

Abstract: Tracking system is a vital aspect of Virtual Reality and Augmented Reality, the efficiency of tracking system is determined by the implementation of framework and the predictive filtering algorithm. As a result of the better applicability of Bayesian predictive filtering algorithm in simulation of non-linear system model, this paper proposes a framework for Bayesian predictive filter, which includes predictive filtering layer and denotation layer, and according to every layer’s function, analyses the implementation of framework. The optimal simulation count is worked out by the experiment. The results show that in the simulation of non-linear system model, this framework for Bayesian predictive filter can implement the tracking of simple motion and the orientation prediction.

Abstract: Integrated Computational Materials Engineering (ICME), and Integrated ComputationalMaterials Science (ICMS) are developing fields with an aim of alloy design, by combining physicalmodels describing materials behavior through lengthscales and processing steps. It has beensuspected, however, that uncertainties in input parameters may cumulate in a hereditary way andyield to a high variability in the final output, independently of the quality of models themselves.Such a variability is however rarely quantified. In this aim, an illustrative example is here given,using a set of “cascade models”, each model being voluntarily very simple (grain growth,precipitation, hardening…) whereas assumed to be exact, so that only the effect of parameteruncertainties on the variability of the output (yield stress of a Ni-base superalloy) can be studied. Itis demonstrated that, with usual uncertainty levels in input parameters, the final dispersion (error)can become very high. Additionally, considering that models are not exact themselves would renderthe situation even worse. Besides, global and implicit models, like neural networks or Gaussianprocesses, have been shown to be able to perform reliable predictions and to be used for alloydesign, with acceptable levels of error, the latter being estimated by statistical methods. In addition,unlike ICME or ICMS, predictions are very fast so that automatic alloy composition optimisation ispossible using, for instance, genetic algorithms. Other fast predictive tools, like computationalthermodynamics (Thermo-Calc), can then be used as constraints during alloy optimisation.

Abstract: In order to carry out effective management on the operation of loom, a new approach is proposed to predict the operating condition of the loom. Firstly, the mathematical basis of Bayesian network is made, so a kind of abbreviated forecasting formula is proposed. Secondly, Bayesian model is made to predict the operating condition of the loom, and the related parameters of the model are estimated by processing real time data in the process of loom production. The practice has shown that the Bayesian prediction model has good results, and Bayesian network is more fit for the operating condition of the loom, which is of great value to judge the operating condition of the loom.

Abstract: Bayesian reinforcement learning has turned out to be an effective solution to the optimal tradeoff between exploration and exploitation. However, in practical applications, the learning parameters with exponential growth are the main impediment for online planning and learning. To overcome this problem, we bring factored representations, model-based learning, and Bayesian reinforcement learning together in a new approach. Firstly, we exploit a factored representation to describe the states to reduce the size of learning parameters, and adopt Bayesian inference method to learn the unknown structure and parameters simultaneously. Then, we use an online point-based value iteration algorithm to plan and learn. The experimental results show that the proposed approach is an effective way for improving the learning efficiency in large-scale state spaces.

Abstract: Erbium-doped fiber source for Fiber Optic Gyro (FOG) uses doped fiber to produce super fluorescence with laser pumping. It has higher output power, wide spectral lines, lower temporal coherence, good temperature stability and long life, which are perfect source for high precision FOG. To solve the problem of reliability analysis of erbium-doped fiber source for FOG with zero failure data, Weibull distribution is chosen as the life distribution model of erbium-doped fiber source on basis of the failure mechanism analysis in this paper. And Bayesian theory is used to estimate the failure rate in different time with zero failure data, then the parameters of the life model are estimated to get reliability index of erbium-doped fiber source. The method greatly decreases the number of test samples because of Bayesian estimation has take advantage of experience information, and also, it overcomes the shortcoming of relying on failure data when using traditional reliability assessment methods. So it has great value on project application.

Abstract: Pollution point source identification for the non-shore emission which is the main form of sudden water pollution incident is considered in this paper. Firstly, the source traceability of sudden water pollution accidents is taken as the Bayesian estimation problem; secondly, the posterior probability distribution of the source's parameters are deduced; thirdly, the marginal posterior probability density is obtained by using a new traceability method; finally, this proposed method is compared with Bayesian-MCMC by numerical experiments. The conclusions are as following: the new traceability method can reduce the iterations, improve the recognition accuracy, and reduce the overall average error obviously and it is more stable and robust than Bayesian-MCMC and can identify sudden water pollution accidents source effectively. Therefore, it provides a new idea and method to solve the difficulty of traceability problems in sudden water pollution accidents.

Abstract: This paper proposes Bayesian statistical method to identify the video traffic by the symmetrical features and coding statistical characteristics of video calls. According to the problem of high computational complexity of the non-parametric probability density estimate method in the condition of large samples, we propose grid probability density estimation method of gird division to reduce the computational complexity. We present identification results. The experimental results indicate that that this method can effectively detect video call traffic.