Supervised classification using probabilistic decision graphs

Abstract

A new model for supervised classification based on probabilistic decision graphs is introduced. A probabilistic decision graph (PDG) is a graphical model that efficiently captures certain context specific independencies that are not easily represented by other graphical models traditionally used for classification, such as the Naïve Bayes (NB) or Classification Trees (CT). This means that the PDG model can capture some distributions using fewer parameters than classical models. Two approaches for constructing a PDG for classification are proposed. The first is to directly construct the model from a dataset of labelled data, while the second is to transform a previously obtained Bayesian classifier into a PDG model that can then be refined. These two approaches are compared with a wide range of classical approaches to the supervised classification problem on a number of both real world databases and artificially generated data.

Modelling and inference with Conditional Gaussian Probabilistic Decision Graphs

“…Assume we want to compile the CG-PDG described there, in order to incorporate evidence (W 2 = 0). The restriction of the model to (W 2 = 0) results in the following changes: f ν 5 (1) = f ν 6 (1) = f ν 7 (1) = 0, and the compilation of the restricted models requires the updating of the following parameters: μ ν 8 = 5.5, s 2 ν 8 = 1.46, μ ν 9 = 8.5 and s 2 ν 9 = 1.9924.…”

“…The PDG model has also been successfully applied to supervised classification problems [5] and unsupervised clustering [6].…”

The construction of a hospital disease tracking and control system with a disease infection probability model

“…Dynamic programming is often adopted to help reduce the time complexity of the algorithm (Cicalese et al 2011). In supervised classification training, for instance, decision graphs have been found to significantly enhance the efficiency of classification models constructed for wireless network planning (Nielsen et al 2009). In clinical and epidemiological analysis, tree diagrams have also been used to formulate classification rules that help clinical practitioners to distinguish between clusters of signs and symptoms and quickly reach a diagnosis (Marshall 2001).…”

Classification Binary Models for Biomedical Data: Simple Probabilistic Networks and Logistic Regression

“…Thus many techniques can be used in the binary classification but methods with high performance are highly required to minimize risks where diagnosis mistakes can cost the life of the patients. In this context, Bayesian networks have emerged as a practical classification technology with successful applications in many fields and provides some advantages such as the ability to combine expert opinion and experimental data (NIELSEN et al, 2009;HECKERMAN et al, 1995).…”