A New Double Weighted Fuzzy Hypergeometric Naive Bayes Network and its Application for User’s Assessment in Virtual Reality Simulators

Abstract

Abstract Assessment methods have been used for identify the quality of procedures performed by human beings. In virtual reality simulators, user’s interaction data can be collected and utilized by Single User Assessment Systems (SUAS) to assess the user’s performance. In particular, some procedures can be considered well performed when the same objective is achieved a number of times with success within a limited universe of attempts. In this case, the data can be modeled by a Hypergeometric distribution. This paper presents the proposal of a new Double Weighted Fuzzy Hypergeometric Naive Bayes (DW-FHyperNB) Network as basis for a SUAS, which can be used in virtual reality simulators. The results showed that this new SUAS was able to achieve better results when compared to other SUAS based on different Naive Bayes Networks. This comparison was performed using SUAS based on Fuzzy Hypergeometric Naive Bayes Network, Classical Hypergeometric Naive Bayes Network, Double Weighted Classical Hypergeometric Naive Bayes Network, Multinomial Naive Bayes Network, Bayes Net, Support Vector Machine, Multilayer Perceptron Neural Network, Radial Basis Network with Gaussian Functions, C4.5 Decision Tree, Decision Table-Naive Bayes, Multinomial Logistic Regression, Simple Cart, and Random Forest was performed. The results obtained showed that the DW-FHyperNB Network produced the best performance, according to the Overall Accuracy Index, Kappa and Tau Coefficients, and diagnostic tests. The new DW-FHyperNB Network can also be utilized for data and image classification, pattern recognition, as well as machine learning applications.

Keywords

• Fuzzy hypergeometric Naive Bayes
• Double weighted fuzzy hypergeometric Naive Bayes
• Hypergeometric distribution
• User’s assessment
• Virtual reality
• Machine learning