Cost-sensitive sequential three-way decision modeling using a deep neural network

Li, Huaxiong; Zhang, Libo; Zhou, Xianzhong; Huang, Bing

doi:10.1016/j.ijar.2017.03.008

Cited by 188 publications

(28 citation statements)

References 42 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…The most meaningful approach to deal with this type of ambiguity, which amounts to a lack of evidence for certain cases or instances, consists in allowing the classifier to abstain, even partially, that is excluding some of the alternative possible classifications. This approach to classification has already been explored in three-way decision theory [6], that recently attracted great interest in the domain of granular computing [9] and Machine Learning [27], possible rules in Rough Set Theory [14] and cautious classifiers [7]. In doing so, there is a trade-off between the coverage of a classifier algorithm, i.e., the instances on which the classifier provides a decision, and its reliability and robustness to classification errors, while striving to keep as large an accuracy as possible: the goal is to learn classifiers that are still as precise as possible, but express a prediction only when they are sufficiently confident.…”

Section: Three-way Outputmentioning

confidence: 99%

The three-way-in and three-way-out framework to treat and exploit ambiguity in data

Campagner

Cabitza

Ciucci

2020

International Journal of Approximate Reasoning

View full text Add to dashboard Cite

In this paper, we address ambiguity, intended as a characteristic of any data expression for which a unique meaning cannot be associated by the computational agent for either lack of information or multiple interpretations of the same configuration. In particular, we will propose and discuss ways in which a decision-support classifier can accept ambiguous data and make some (informative) value out of them for the decision maker. Towards this goal we propose a set of learning algorithms within what we call the three-way-in and three-way-out approach, that is, respectively, learning from partially labeled data and learning classifiers that can abstain. This approach is based on orthopartitions, as a common representation framework, and on three-way decisions and evidence theory, as tools to enable uncertain and approximate reasoning and inference. For both the above learning settings, we provide experimental results and comparisons with standard Machine Learning techniques, and show the advantages and promising performances of the proposed approaches on a collection of benchmarks, including a real-world medical dataset. a logic of incomplete information, cast in the setting of possibility theory.

show abstract

Section: Three-way Outputmentioning

confidence: 99%

The three-way-in and three-way-out framework to treat and exploit ambiguity in data

Campagner

Cabitza

Ciucci

2020

International Journal of Approximate Reasoning

View full text Add to dashboard Cite

show abstract

“…In light of Bayesian decision procedure [4], decision theoretic rough set (DTRS) model was proposed by Yao and Wong [35] to analyze the noisy data by considering the tolerance of classification error. Since then the DTRS model has found its applications in various theoretical and practical fields, and it has produced many god results [2,[15][16][17]20,25].…”

Section: Introductionmentioning

confidence: 99%

Multi-granulation Pythagorean fuzzy decision-theoretic rough sets based on inclusion measure and their application in incomplete multi-source information systems

Mandal¹,

Ranadive

2019

Complex Intell. Syst.

View full text Add to dashboard Cite

Multi-granulation rough sets (MGRSs) and decision-theoretic rough sets (DTRSs) are two important and popular generalizations of classical rough sets. The combination of two generalized rough sets have been investigated by numerous researchers in different extensions of fuzzy settings such as interval-valued fuzzy sets (IVFSs), intuitionistic fuzzy sets (IFSs), bipolar-valued fuzzy sets (BVFSs), etc. Pythagorean fuzzy (PF) set is another extension of fuzzy set, which is more capable in comparison to IFS handle vagueness in real world. However, few studies have focused on the combination of the two rough sets in PF settings. In this study, we combine the two generalized rough sets in PF settings. First, we introduce a type of PF subset (of subset of the given universe) of the PF Set (of the given universe). Then we establish two basic models of multi-granulation PF DTRS (MG-PF-DTRS) of PF subset of the PF set based on PF inclusion measure within the framework of multi-granulation PF approximation space. One model is based on a combination of PF relations (PFRs) and the construction of approximations with respect to the combined PFR. By combining PFRs through intersection and union, respectively, we construct two models. The other model is based on the construction of approximations from PFRs and a combination of the approximations. By using intersection and union to combine the approximations, respectively, we again get two models. As a result, we have total four models. Further for different constraints on parameters, we obtain three kinds of each model of the MG-PF-DTRSs. Then, their principal structure, basic properties and uncertainty measure methods are investigated as well. Second, we give a way to compute PF similarity degrees between two objects and also give a way to compute PF decision-making objects from incomplete multi-source information systems (IMSISs). Then we design an algorithm for decision-making to IMSISs using MG-PFDTRSs and their uncertainty measure methods. Finally, an example about the mutual funds investment is included to show the feasibility and potential of the theoretic results obtained.

show abstract

“…Cabitza et al [3] presented two methods based on three-way decisions for collective knowledge extraction from questionnaires. By considering the misclassification cost and test cost, Li et al [14] proposed a cost-sensitive sequential 3WD strategy to analyze image data. In addition, theoretic researches have also achieved many fruits [5], [6], [18], [22], [31], [46], [50]- [52].…”

Section: Introductionmentioning

confidence: 99%

Hierarchical Three-Way Decisions With Intuitionistic Fuzzy Numbers in Multi-Granularity Spaces

2019

View full text Add to dashboard Cite

Intuitionistic fuzzy sets, as extensions of fuzzy sets, are described by intuitionistic fuzzy numbers (IFNs) that describe the uncertain concepts with the membership and non-membership degrees together. Three-way decisions are effective classification methods, which are always utilized for solving uncertain problems with fixed cost parameter values and fixed attribute values in reality. However, under dynamic intuitionistic fuzzy environments, three-way decisions face a great challenge for processing uncertain problems with IFN cost parameters and IFN attribute values. In this paper, by considering the impact of IFNs, a hierarchical three-way decision model with IFN cost parameters (H3WDIF-I) is established in multi-granularity spaces for handling dynamic and uncertain problems first. The change rules of H3WDIF-I are discussed to analyze the relationships of decisions based on the granularity structures. In addition, when the IFN cost parameters and IFN attribute values arise together, a hierarchical three-way decision model with IFNs (H3WDIF-II) is proposed to explore three-way decisions in depth under dynamic intuitionistic fuzzy environments. Similarly, the change rules of H3WDIF-II are also discussed from the viewpoint of classification losses in multi-granularity spaces. Finally, the presented change rules are verified by many examples and experiments.

show abstract

Cost-sensitive sequential three-way decision modeling using a deep neural network

Cited by 188 publications

References 42 publications

The three-way-in and three-way-out framework to treat and exploit ambiguity in data

The three-way-in and three-way-out framework to treat and exploit ambiguity in data

Multi-granulation Pythagorean fuzzy decision-theoretic rough sets based on inclusion measure and their application in incomplete multi-source information systems

Hierarchical Three-Way Decisions With Intuitionistic Fuzzy Numbers in Multi-Granularity Spaces

Contact Info

Product

Resources

About