China
Africa
Explore chapters and articles related to this topic
Impact of Ensemble-Based Models on Cancer Classification, Its Development, and Challenges
Published in Om Prakash Jena, Bharat Bhushan, Nitin Rakesh, Parma Nand Astya, Yousef Farhaoui, Machine Learning and Deep Learning in Efficacy Improvement of Healthcare Systems, 2022
Barnali Sahu, Sitarashmi Sahu, Om Prakash Jena
Naïve-Bayes (NB) classification algorithm works on the basis of Bayes' conditional probability theorem. In this, probability means the degree of belief. The conditional probability is being used to classify the data. The most important part of this algorithm is that it works with the assumption that all of the attributes are independent of each other. There are three different kinds of NB-based algorithms present as the Gaussian NB, multinomial NB, and Bernoulli NB. The main advantage present behind this classification is that it requires a very small amount of training data for estimating the conditional parameters but the estimation time depends upon the data set size. If the dimension of the data set is very high, then the NB will be acting as a bad estimator as the estimation time and cost increases concerning the data set.
Characterizing Uncertainty through Expert Elicitation
Published in Charles Yoe, Principles of Risk Analysis, 2019
Risk managers frequently must ask questions which science alone is incapable of answering. For example, what will happen to accident rates when autonomous vehicles comprise a significant part of the traffic? How will sea level change affect the fortunes of a particular East coast port in this century? A subjective or degree-of-belief approach to probability is useful for many of these situations. We all have gotten quite used to making informal probabilistic judgments about the uncertainty in our own lives with this degree-of-belief approach and we do this with surprising ease and regularity. You don't take an umbrella if you believe it will not rain hard enough to need one. You cross the street if you believe you will not get hit. When you handicap your favorite team's chances to win their next game or the incumbent party's likelihood to retain the White House, you are addressing uncertainty through the subjective assessment of probability. Over time you learn where and when your instincts are more or less good, and you make decisions accordingly.
Approach to a Bayesian decision model for cost-benefit analysis in security risk
Published in Stein Haugen, Anne Barros, Coen van Gulijk, Trond Kongsvik, Jan Erik Vinnem, Safety and Reliability – Safe Societies in a Changing World, 2018
Bayesian Networks (BN) are based on Bayesian probabilistics interpreting probability as a degree of belief. BN represent a combination of probability and graph theory. A BN therefore quantifies dependencies between various data, information or knowledge considering uncertainties (Jensen & Nielsen 2007). BN consist of nodes and connecting edges in directed acyclic graphs (DAG) linking parent and children nodes. BN therefore consist of (Gribaudo et al. 2015): Variables (nodes) with a finite set of statesDirected edges between the nodesA conditional probability table describing the result of each node
Bayesian Network–Based Fault Diagnostic System for Nuclear Power Plant Assets
Published in Nuclear Technology, 2023
Xingang Zhao, Xinyan Wang, Michael W. Golay
The BN, also known as the Bayesian belief network, Bayesian net, or causal probabilistic network, is a probabilistic graphical model built upon the Bayes’ theorem that interprets probability as the statement of a degree of belief in the occurrence of any event based upon prior and observed evidence. A BN is represented by a directed acyclic graph made of a set of variable nodes and directed edges/links symbolizing the causal relations between variables. Each variable node, whether discrete or continuous,aFor continuous variable nodes, distribution functions or variable discretization are needed in existing BN software tools.17 has a finite set of mutually exclusive, collectively exhaustive (MECE) states. The strength of the causal relations between variables is quantified in the conditional probability table (CPT) attached to each linked node. A CPT specifies the degree of belief that the corresponding node will be in a particular state given the states of its parent/causal node(s).bThe root (or parentless) nodes are described by marginal (or prior) probability distributions rather than conditional probabilities. For each child/effect node , the attached CPT is
Parametric empirical Bayes estimation of individual time-pressure reactivity
Published in International Journal of Production Research, 2018
The prior distribution represents the degree of belief about the state of nature before individual-specific data are available. In general, the form of the prior distribution is not pre-determined, but rather it depends upon context and the availability of related knowledge. However, for certain specific prior forms (conjugate priors), Bayesian estimation can be done analytically in closed form rather than numerically. The conjugate prior is a parametric distribution that is mathematically similar to the likelihood, and it yields a form parametrically similar to the posterior distribution (Wilks 2011). For the gamma likelihood, as used in this study, the conjugate prior is also a gamma distribution (Bolstad 2010; Luo and Altman 2013); The conjugate prior for k is known as Gamma (α, β), as shown in Equation (5). In Equation (5), k represents the time-pressure reactivity that was used in Equation (3). α is a shape parameter of the prior gamma distribution and β is the reciprocal of the scale parameter.
Identification of China’s strategic transport passages in the context of the Belt and Road initiative
Published in Maritime Policy & Management, 2023
Daozheng Huang, Sean Loughney, Jin Wang
The ER approach offers a rational and reproducible methodology to aggregate uncertain, incomplete, and vague data. ER uses the concept of ‘degree of belief’ to elicit a decision-maker’s preference. The degree of belief can be described as the degree of expectation that an alternative will yield an anticipated outcome on a particular criterion. An individual’s degree of belief depends on the knowledge of the subject and the experience (Wang, Yang, and Sen 1995; Yang and Xu 2002; Sadeghi et al. 2018). The ER approach has been developed particularly for multiple attribute decision making problems with both qualitative and quantitative criteria under uncertainties utilizes individuals’ knowledge, expertise, and experience in the forms of belief functions. The major advantage of ER is its ability to handle incomplete, uncertain, and vague as well as complete and precise data. However, there are two quantitative parts to ER, one is the belief degrees, and the other is the relative weights of the criteria. Analytical Hierarchy Process (AHP) is an ideal solution to develop these weights as the data gathering process can incorporate both the belief degree determination and Pairwise Comparisons (PCs), which is a tremendous advantage in the data gathering process. This is particularly true when utilizing non-probability sampling, as it allows experts to complete the surveys for both ER and AHP at the same time, thus, limiting the level of uncertainty and randomness related to separate surveys for other mixed approaches (Sönmez et al. 2012). Thus, the ER approach, combined with AHP is ideal for application to the assessment of China’s STPs.