Density maximisation classification in the lattice machine

This paper reviews the lattice machine classification framework (H. Wang et al., 1999), (Hui Wang et al., 2000), (Hui Wang et al., 2004) and its classification methods, in particular the density maximisation method (Hui Wang et al., 2003). This paper also suggests a different way of estimating density, which is based on the contextual probability (H. Wang and W. Dubitzky, 2005). The lattice machine approximates data resulting in, as a model of data, a set of hypertuples that are equilabelled, supported and maximal. Such a model can be used for classification with the C2 method (Hui Wang et al., 2000) or the density maximisation method (Hui Wang et al., 2003). The density maximisation method uses the lattice machine model of data to classify new data with a view to maximising the density of the model. The density maximisation method uses a simple definition of density. In this paper we suggest using a different density estimation method, which is based on the contextual probability