Tree-structured Bayesian Networks Research Articles

Bayesian networks + reinforcement learning: Controlling group emotion from sensory stimuli

As communication technology develops, various sensory stimuli can be collected in service spaces. To enhance the service effectiveness, it is important to determine the optimal stimuli to induce group emotion in the service space to the target emotion. In this paper, we propose a stimuli control system to adjust the group emotion. It is a stand-alone system that can determine optimal stimuli by utility table and modular tree-structured Bayesian networks designed for emotion prediction model proposed in the previous study. To verify the proposed system, we collected data using several scenarios at a kindergarten and a senior welfare center. Each space is equipped with sensors for collection and equipment for controlling stimuli. As a result, the system shows a performance of 78% in the kindergarten and 80% in the senior welfare center. The proposed method shows much better performance than other classification methods with lower complexity. Also, reinforcement learning is applied to improving the accuracy of stimuli decision for a positive effect on system performance.

A circular-linear dependence measure under Johnson–Wehrly distributions and its application in Bayesian networks

Circular data jointly observed with linear data are common in various disciplines. Since circular data require different techniques than linear data, it is often misleading to use usual dependence measures for joint data of circular and linear observations. Moreover, although a mutual information measure between circular variables exists, the measure has drawbacks in that it is defined only for a bivariate extension of the wrapped Cauchy distribution and has to be approximated using numerical methods. In this paper, we introduce two measures of dependence, namely, (i) circular-linear mutual information as a measure of dependence between circular and linear variables and (ii) circular-circular mutual information as a measure of dependence between two circular variables. It is shown that the expression for the proposed circular-linear mutual information can be greatly simplified for a subfamily of Johnson–Wehrly distributions. We apply these two dependence measures to learn a circular-linear tree-structured Bayesian network that combines circular and linear variables. To illustrate and evaluate our proposal, we perform experiments with simulated data. We also use a real meteorological data set from different European stations to create a circular-linear tree-structured Bayesian network model.

A Mixtures-of-Trees Framework for Multi-Label Classification.

We propose a new probabilistic approach for multi-label classification that aims to represent the class posterior distribution P(Y|X). Our approach uses a mixture of tree-structured Bayesian networks, which can leverage the computational advantages of conditional tree-structured models and the abilities of mixtures to compensate for tree-structured restrictions. We develop algorithms for learning the model from data and for performing multi-label predictions using the learned model. Experiments on multiple datasets demonstrate that our approach outperforms several state-of-the-art multi-label classification methods.

Kazakh Traditional Dance Gesture Recognition

Full body gesture recognition is an important and interdisciplinary research field which is widely used in many application spheres including dance gesture recognition. The rapid growth of technology in recent years brought a lot of contribution in this domain. However it is still challenging task. In this paper we implement Kazakh traditional dance gesture recognition. We use Microsoft Kinect camera to obtain human skeleton and depth information. Then we apply tree-structured Bayesian network and Expectation Maximization algorithm with K-means clustering to calculate conditional linear Gaussians for classifying poses. And finally we use Hidden Markov Model to detect dance gestures. Our main contribution is that we extend Kinect skeleton by adding headwear as a new skeleton joint which is calculated from depth image. This novelty allows us to significantly improve the accuracy of head gesture recognition of a dancer which in turn plays considerable role in whole body gesture recognition. Experimental results show the efficiency of the proposed method and that its performance is comparable to the state-of-the-art system performances.

Deformable Bayesian Network: A Robust Framework for Underwater Sensor Fusion

The dynamic tree (DT) graphical model is a popular analytical tool for image segmentation and object classification tasks. A DT is a useful model in this context because its hierarchical property enables the user to examine information in multiple scales and its flexible structure can more easily fit complex region boundaries compared to rigid quadtree structures such as tree-structured Bayesian networks. This paper proposes a novel framework for data fusion called a deformable Bayesian network (DFBN) by using a DT model to fuse measurements from multiple sensing platforms into a nonredundant representation. The structural flexibility of the DFBN will be used to fuse common information across different sensor measurements. The appropriate structure update strategies for the DFBN and its parameters for the data fusion application are discussed. A real-world example application using sonar images collected from a survey mission is presented. The fusion results using the presented DFBN framework are shown to outperform state-of-the-art approaches such as the Gaussian mean shift and spectral clustering algorithms. The DFBN's complexity and scalability are discussed to address its potential for a larger data set.

Tree-structured Bayesian network learning with application to scene classification

A tree-structured Bayesian network is one of the best probabilistic models for scene classification. A simple and successful learning algorithm for a tree-structured Bayesian network classifier is presented without taking arc reversal into account. A systematic performance study on the Lazebnik 15 dataset shows the proposed method is more efficient than the traditional learning scheme.

NML Computation Algorithms for Tree-Structured Multinomial Bayesian Networks

Typical problems in bioinformatics involve large discrete datasets. Therefore, in order to apply statistical methods in such domains, it is important to develop efficient algorithms suitable for discrete data. The minimum description length (MDL) principle is a theoretically well-founded, general framework for performing statistical inference. The mathematical formalization of MDL is based on the normalized maximum likelihood (NML) distribution, which has several desirable theoretical properties. In the case of discrete data, straightforward computation of the NML distribution requires exponential time with respect to the sample size, since the definition involves a sum over all the possible data samples of a fixed size. In this paper, we first review some existing algorithms for efficient NML computation in the case of multinomial and naive Bayes model families. Then we proceed by extending these algorithms to more complex, tree-structured Bayesian networks.

Effective Dimensions of Hierarchical Latent Class Models

Hierarchical latent class (HLC) models are tree-structured Bayesian networks where leaf nodes are observed while internal nodes are latent. There are no theoretically well justified model selection criteria for HLC models in particular and Bayesian networks with latent nodes in general. Nonetheless, empirical studies suggest that the BIC score is a reasonable criterion to use in practice for learning HLC models. Empirical studies also suggest that sometimes model selection can be improved if standard model dimension is replaced with effective model dimension in the penalty term of the BIC score. Effective dimensions are difficult to compute. In this paper, we prove a theorem that relates the effective dimension of an HLC model to the effective dimensions of a number of latent class models. The theorem makes it computationally feasible to compute the effective dimensions of large HLC models. The theorem can also be used to compute the effective dimensions of general tree models.

The automated extraction of environmentally relevant features from digital imagery using Bayesian multi-resolution analysis

In this paper, we discuss the use of hierarchical tree-structured Bayesian networks for integrating knowledge concerning contextual relationships between environmentally relevant features extracted from digital imagery at multiple resolution scales. In our model, conditional probability distributions over continuous valued observations are parameterized using a mixture of multivariate Gaussian distributions. Separate classifiers for pixels and groups of pixels are used as sub-components of the overall model. The Bayesian formalism allows models to be composed in a systematic and statistically sound manner. We illustrate how this approach can be used to resolve ambiguity leading to classification errors and thus improve techniques for the classification of land use from aerial imagery. We present an example relevant to ecosystem analysis, the monitoring of urban growth and the automatic generation of input parameters for hydrologic models.