Qualitative probabilistic networks with reduced ambiguities

Abstract

A Qualitative probabilistic network (QPN) is the qualitative abstraction of a Bayesian network that encodes variables and the qualitative influences between them. In order to make QPNs be practical for real-world representation and inference of uncertain knowledge, it is desirable to reduce ambiguities in general QPNs, including unknown qualitative influences and inference conflicts. In this paper, we first extend the traditional definition of qualitative influences by adopting the probabilistic threshold. In addition, we introduce probabilistic-rough-set-based weights to the qualitative influences. The enhanced network so obtained, called EQPN, is constructed from sample data. Finally, to achieve conflict-free EQPN inferences, we resolve the trade-offs by addressing the symmetry, transitivity and composition properties. Preliminary experiments verify the correctness and feasibility of our methods.

Integration of Multiple Temporal Qualitative Probabilistic Networks in Time Series Environments

“…According to Pawlak's rough set theory, the zero dependency degree can be associated with the qualitative influence whose strength should not be zero actually [15], so Yue et al [15] adopted the probabilistic rough set theory to obtain dependency degree as the strength between the associated variables.…”

“…It represents a very special type of similarity between elements of the universe. According to [13][14][15], we redescribe the following three definitions. / R Definition 3 Let X be a set of objects in U and R be an equivalence relation over .…”

“…To reduce the ambiguities, Yue et al [15] introduced the rough-set-based dependency degree as the strength of qualitative influence. We use the following Example 1 to illustrate the rough-set-based dependency degree.…”

A Modeling and Probabilistic Reasoning Method of Dynamic Uncertain Causality Graph for Industrial Fault Diagnosis

“…The complexity becomes particularly problematic for large and multiconnected network [17] . Therefore, some studies on efficient inference algorithms and the structure compilation or conversions to BNs have been emerging, such as variable elimination [18] , recursive conditioning algorithms [19] , the enhanced qualitative probabilistic network [20] , the decomposable negation normal form [21] , multiplicative factorization for the noisy-MAX [22] , weighted CNF encoding algorithm [23] , the qualitative characterization method of ICI models [24] , etc. However, challenges also lie in the insufficiency of historical fault dada.…”

QMIQPN: An enhanced QPN based on qualitative mutual information for reducing ambiguity

“…Furthermore, reasoning with qualitative probabilities is much more efficient than reasoning with precise ones, the inference complexity of QPN is a polynomial in the size of the network [5], rather than NP-hard [6]. Therefore, many approaches have been proposed for QPN modeling and inference according to various kinds of applications [7][8][9][10][11][12][13][14][15][16][17][18][19][20].…”