An improvement for combination rule in evidence theory

2019

The likelihood function is widely used in data processing but the classical likelihood function is too strict to deal with data with extreme values in real applications.To address this issue, a soft likelihood function based on power ordered weighted average (POWA) operator for probability distribution is proposed. Compared with the existing soft likelihood functions, one of the main features of the proposed likelihood function is that the weighted mean is displayed by the geometric mean of the weighted probability products. As a result, the effect on final result caused by extreme data can be efficiently decreased. Numerical examples are used to illustrate the efficiency of the proposed soft likelihood function. K E Y W O R D S geometric mean, likelihood function, OWA operator, POWA operator, soft likelihood function | INTRODUCTIONDealing with uncertainty information plays an important role in real applications. 1-3 Many math tools, such as fuzzy sets, 4-6 rough sets, 7 Z-numbers, 8-11 belief structures, 12-14 D numbers 15-17 entropy function, 18,19 and evidence theory, 20-22 are presented to handle different types of uncertainty information. Among these tools, probability is heavily studied and is widely used in many applications, including decision making, 23-26 estimation, 27,28 and uncertain measurement. [29][30][31][32] One of the most used models in probability is the likelihood function, which is used when data are available to describe plausibility of a parameter value. [33][34][35] However, it is very strict in real applications since the original likelihood function made up with the probability product is too strict to show the compatibility probabilities for the different pieces of evidence since any piece of probabilities, which are low-value would greatly reduce the likelihood function.How to cite this article: Song Y, Deng Y. A new soft likelihood function based on power ordered weighted average operator. Int J Intell

Section: Preliminariesmentioning

confidence: 99%

A new soft likelihood function based on power ordered weighted average operator

Song

2019

“…Besides, Su et al proposed a model based on the Pearson correlation coefficient to measure the dependence of the evidence. Some researchers also identified independence in Bayesian networks whereas others's idea is to measure the degree of conflict and define the distance or the dissimilarity between evidence considering combining the evidence to address conflict . Recently, a method based on shearman coefficient to combine dependent is presented .…”

Section: Introductionmentioning

confidence: 99%

“…Some researchers also identified independence in Bayesian networks whereas others's idea is to measure the degree of conflict and define the distance 33 or the dissimilarity between evidence 34 considering combining the evidence to address conflict. [35][36][37][38] Recently, a method based on shearman coefficient to combine dependent is presented. 39 However, the indirect relationship among evidence is neglected to some degree.…”

mentioning

confidence: 99%

Dependent evidence combination based on decision‐making trial and evaluation laboratory method

2019

Dempster‐Shafer is widely used to address the problems of uncertainty. One assumption mentioned in this theory is that the distribution of information should be independent. In practice, the requirement cannot be fulfilled. One of the efficient methods to deal with dependent evidence is to calculate the correlation discounting. However, existing coefficient can only be applied to show the direct relation between evidence A and B but do not take the indirect relationship into consideration. To address this issue, in this paper, a new method to combine dependent evidence based on decision‐making trial and evaluation laboratory is presented, not only considering the relation between evidence A and B and the relation between evidence B and C, but also considering the transitive influence between evidence A and C. Finally, the experiments on some benchmark data sets are illustrated to show the efficiency of the proposed method.

“…In addition, Dempster's combination rule can combine evidence from different sources without prior information . However, it should be pointed out that illogical results may be obtained by classical Dempster combination rule when collected evidence highly conflicts each other …”

Section: Introductionmentioning

confidence: 99%

TDBF: Two‐dimensional belief function

2019

How to efficiently handle uncertain information is still an open issue. In this paper, a new method to deal with uncertain information, named as two-dimensional belief function (TDBF), is presented. A TDBF has two components, T = (m m , A B ), both m A and m B are classical belief functions, while m B is a measure of reliable of m A . The definition of TDBF and the discounting algorithm are proposed. Compared with the classical discounting model, the proposed TDBF is more flexible and reasonable. Numerical examples are used to show the efficiency and application of the proposed method. K E Y W O R D S belief function, conflict management, Dempster-Shafer evidence theory, information fusion, two-dimensional belief function, Z-numbers 1 | INTRODUCTIONIt is inevitable to deal with uncertain information in real word. 1-3 There are many articles that take advantage of uncertain information processing problems. 4 To address this issue, many methods have been proposed, such as probability theory, 5 Dempster-Shafer evidence theory, 6,7 fuzzy sets, 8-10 rough sets, 11 Z-numbers, 12 R-numbers, 13,14 and D numbers. [15][16][17][18][19][20] Among these methods, Dempster-Shafer evidence theory 6,7 is one of the most widely used math tools in many engineering applications such as pattern recognition 21,22 and decision-making. 23 Evidence theory has many advantages to handle uncertain information. [24][25][26] For example, the belief function is more flexible to model uncertainty compared with probability distribution. In addition, Dempster's combination rule can combine evidence from different sources without prior information. 27,28 However, it should be pointed out that illogical results may be obtained by classical Dempster combination rule when collected evidence highly conflicts each other. [29][30][31][32][33] Z-numbers is proposed by Zadeh. 12 A Z-number has two components, Z = (A, B). The first component, A, is a restriction (constraint) on the values which a real-valued uncertain variable, X, is allowed to take. The second component, B, is a measure of reliability (certainty) of the first component. 12 Similarly, in evidence theory, discounting coefficient is a measure of sensor report's reliability, for example, discounting factors based on dissimilarity measure, 34 dynamic discounting rates. 35 Nevertheless, the simple value is not enough and reasonable to express the experts evaluation under uncertain situation. To solve this problem, a new math model named as two-dimensional belief function (TDBF) is proposed in this paper.A TDBF is an ordered pair of basic probability assignments (BPAs) denoted as T = (m m , A B ). The first component m A is a classical BPA, which usually is collected from the sensors. The second component m B is a measure of reliability for the first component, which can be collected by the experts. The frame of discernment of m B set Y N Θ = { , }, where "Y" denotes "positive" and "N" denotes "negative." The power set of Y N Θ = { , } consists of two singletons {Y} and {N}, an universal Θ and t...