A New Soft Likelihood Function Based on D Numbers in Handling Uncertain Information

Tian, Ye; Mi, Xiangjun; Liu, Lili; Kang, Bingyi

doi:10.1007/s40815-020-00911-z

Cited by 25 publications

(9 citation statements)

References 62 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…From the fusion results of 𝑁 pieces of evidence using equation (15), it can be seen that after each fusion, for the 𝑖th evidence, factorization associated with the order 𝑖 of evidence includes following two multiplication equations ( 16) and equations (17).…”

Section: Algorithmmentioning

confidence: 99%

“…For the sake of solving this problem, many math theories are talked over and applied to treat indeterminate, fuzzy as well as inaccurate sensor data. These theories contain Bayesian principle [11], fuzzy set principle [12], Rough set [13], evidence theory [14], evidential reasoning [15], Z number [16], D number [17], etc [18][19][20]. In this study, we mainly study the evidence theory was on behalf of belief function to treat multi-sensor information fusion.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

A Weighted Evidence Combination Method for Multisensor Data Fusion

Liu¹,

Liu²

2022

Journal of Internet Technology

View full text Add to dashboard Cite

<p>Multisensor information fusion exerts a key part in lots of practical usages. Dempster-shafer evidence theory has drawn extensive attention in many scopes of information fusion due to its flexibility and effectiveness in dealing with uncertain data without aforehand data. But when combining highly contradictory evidence with Dempster’s combinatorial principles, it can result in counterintuitive results. To solve the issue, the study proposes a multi-sensor data weighted evidence combination fusion method based on inter-evidence difference measure. Firstly, different measures including evidence distance and conflict are with the definition of characterizing distinctions between the two pieces of evidence. Then, according to the difference between each evidence and the average evidence, the weight coefficients of each evidence are calculated. In the end, initial evidence is discounted according to weighting factor, as well as Dempster’s combination principle is adopted to discount the evidence for fusion. Many instances show that this way can efficiently treat highly conflicting evidence and has good convergence performance.</p> <p> </p>

show abstract

Section: Algorithmmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

A Weighted Evidence Combination Method for Multisensor Data Fusion

Liu¹,

Liu²

2022

Journal of Internet Technology

View full text Add to dashboard Cite

show abstract

“…D-S Evidence theory also proposes a combination rule for multi-source information, the Dempster combination rule [19,20,21]. There are many directions for the expansion of evidence theory: D numbers [22,23], evidential modeling [24,25], rule-based [26,27,28], evidence theory and fuzzy set [29,30], evidential neural network [31,32], and complex evidence theory [33], and others [34], which has extensions in different fields [35,36,37]. More combination methods also appeared in the later period of combination rules.…”

Section: Preliminariesmentioning

confidence: 99%

A Fast Evidential Approach for Stock Forecasting

Zhan,

Xiao

2021

Preprint

View full text Add to dashboard Cite

In the framework of evidence theory, the confidence functions of different information can be combined to a combined confidence function. Stock price prediction is the focus of economics. Stock price forecasts can provide reference data. The Dempster combination rule is a classic method of fusing different information. By using the Dempster combination rule and confidence function based on the entire time series fused at each time point and future time points, and the preliminary forecast value obtained through the time relationship, the accurate forecast value can be restored. This article will introduce the prediction method inspired by evidence theory. This method has good running performance, can make a rapid response on a large amount of stock price data, and has far-reaching significance.

show abstract

“…The basic idea of case retrieval by the proposed method is as follows: Firstly, calculate the local similarity between different attributes of the target case and the source case; Then, the CBR-SLFs algorithm proposed in this paper is used to calculate the overall similarity, and some potential available source cases with high similarity are obtained; Finally, the source case solution that is closest to the target case is obtained through KNN, and reuse it. This strategy is developed as a flexible computation of likelihood functions of global similarity calculation, and has the advantage of being more robust and practical in case retrieval [41]. Furthermore, SLF-based case retrieval algorithm is developed introducing an attitudinal characteristic to reflect the subjective preference of decision makers, which allows for more flexible choices based on different types of decision makers.…”

Section: Introductionmentioning

confidence: 99%

A hybrid retrieval strategy for case-based reasoning using soft likelihood functions

et al. 2022

View full text Add to dashboard Cite

Case-based reasoning (CBR) is the retrieval of one or more similar cases from an existing case base for the problem to be solved according to the characteristics of the new problem. The core idea of CBR is that similar cases have similar solutions, so whether the CBR system can play a powerful advantage depends on the quality of case retrieval strategy. At present, the commonly used case retrieval algorithm is based on the mean operator method, which is very hard, and a certain local similarity is low will affect the overall result. In order to calculate the global similarity of cases from a new and softer point of view, this paper introduces the soft likelihood functions into case retrieval, combines the soft likelihood functions with KNN, and proposes a hybrid retrieval strategy.The core of the retrieval strategy is to define the global similarity through SLFs, aggregate the local similarity and characteristic similarity together, and also take the attitude characteristics of decision makers into consideration. Through simulation experiments on real data sets, the accuracy rate is more than 81%, which verifies the effectiveness of the retrieval strategy.

show abstract

A New Soft Likelihood Function Based on D Numbers in Handling Uncertain Information

Cited by 25 publications

References 62 publications

A Weighted Evidence Combination Method for Multisensor Data Fusion

A Weighted Evidence Combination Method for Multisensor Data Fusion

A Fast Evidential Approach for Stock Forecasting

A hybrid retrieval strategy for case-based reasoning using soft likelihood functions

Contact Info

Product

Resources

About