Rapid spline-based kernel density estimation for Bayesian networks

Gurwicz, Yaniv; Lerner, Boaz

doi:10.1109/eeei.2004.1361149

Cited by 4 publications

(4 citation statements)

References 5 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The weight is the statistical measure of correlation, denoted I, and calculated by the statistical independence test that is used by the baseline causal discovery algorithm. For example, mutual information for discrete variables and correlation coefficient for continuous variables (with a rapid density estimation, e.g., using Gurwicz and Lerner [2004]). Clustering can then be viewed as partitioning U into disjoint sub-graphs U 1 , .…”

Section: Domain Variable Clusteringmentioning

confidence: 99%

Improving Efficiency and Accuracy of Causal Discovery Using a Hierarchical Wrapper

Nisimov,

Gurwicz,

Rohekar

et al. 2021

Preprint

View full text Add to dashboard Cite

Causal discovery from observational data is an important tool in many branches of science. Under certain assumptions it allows scientists to explain phenomena, predict, and make decisions. In the large sample limit, sound and complete causal discovery algorithms have been previously introduced, where a directed acyclic graph (DAG), or its equivalence class, representing causal relations is searched. However, in real-world cases, only finite training data is available, which limits the power of statistical tests used by these algorithms, leading to errors in the inferred causal model. This is commonly addressed by devising a strategy for using as few as possible statistical tests. In this paper, we introduce such a strategy in the form of a recursive wrapper for existing constraint-based causal discovery algorithms, which preserves soundness and completeness. It recursively clusters the observed variables using the normalized min-cut criterion from the outset, and uses a baseline causal discovery algorithm during backtracking for learning local sub-graphs. It then combines them and ensures completeness. By an ablation study, using synthetic data, and by common real-world benchmarks, we demonstrate that our approach requires significantly fewer statistical tests, learns more accurate graphs, and requires shorter run-times than the baseline algorithm.

show abstract

Section: Domain Variable Clusteringmentioning

confidence: 99%

Improving Efficiency and Accuracy of Causal Discovery Using a Hierarchical Wrapper

Nisimov,

Gurwicz,

Rohekar

et al. 2021

Preprint

View full text Add to dashboard Cite

show abstract

“…KDE suffers from computational costs making it impractical in many real-world applications. Smooths the density using splines so that it requires fewer coefficients for estimation than the entire training set [19]. The possible continuous feature patterns required for probabilistic inference in a Bayesian network classifier can be calculated by kernel density estimation, letting each pattern influence the shape of the probability density.…”

Section: Naive Bayes Kernelmentioning

confidence: 99%

Improvement of Accuracy and Handling of Missing Value Data in the Naive Bayes Kernel Algorithm

Bijanto

Yunus²

2021

JAIS

View full text Add to dashboard Cite

The lost impact on the research process, can be serious in classifying results leading to biased parameter estimates, statistical information, decreased quality, increased standard error, and weak generalization of the findings. In this paper, we discuss the problems that exist in one of the algorithms, namely the Naive Bayes Kernel algorithm. The Naive Bayes kernel algorithm has the disadvantage of not being able to process data with the mission value. Therefore, in order to process missing value data, there is one method that we propose to overcome, namely using the mean imputation method. The data we use is public data from UCI, namely the HCV (Hepatisis C Virus) dataset. The input method used to correct the missing data so that it can be filled with the average value of the existing data. Before the imputation process means, the dataset uses yahoo bootstrap first. The data that has been corrected using the mean imputation method has just been processed using the Naive Bayes Kernel Algorithm. From the results of the research tests that have been carried out, it can be obtained an accuracy value of 96.05% and the speed of the data computing process with 1 second.

show abstract

“…So, the distribution described in figure 4 cannot be directly model in the network. Some solutions have been proposed to handle various distributions in a Bayesian network: mixture of truncated exponentials ( [21]) or kernels methods ( [22]). But these types of methods implicate some difficulties for the exact inference, and software are not really exploitable.…”

Section: A Presentation Of the Non-gaussian Casementioning

confidence: 99%

Fault detection of univariate non-Gaussian data with Bayesian network

Verron¹,

Tiplica²,

Kobi³

2010

2010 IEEE International Conference on Industrial Technology

View full text Add to dashboard Cite

The purpose of this article is to present a new method for fault detection with Bayesian network. The interest of this method is to propose a new structure of Bayesian network allowing to detect a fault in the case of a non-Gaussian signal. For that, a structure based on Gaussian mixture model is proposed. This particular structure allows to take into account the non-normality of the data. The effectiveness of the method is illustrated on a simple process corrupted by different faults.

show abstract

Rapid spline-based kernel density estimation for Bayesian networks

Cited by 4 publications

References 5 publications

Improving Efficiency and Accuracy of Causal Discovery Using a Hierarchical Wrapper

Improving Efficiency and Accuracy of Causal Discovery Using a Hierarchical Wrapper

Improvement of Accuracy and Handling of Missing Value Data in the Naive Bayes Kernel Algorithm

Fault detection of univariate non-Gaussian data with Bayesian network

Contact Info

Product

Resources

About