Figure 1
From: Classification of heterogeneous microarray data by maximum entropy kernel
Maximum entropy kernel for heterogeneous data. Samples and their distance constraints in the feature space are drawn schematically as graph nodes and edges, respectively. (a) The heterogeneous data are entangled in the feature space, making it difficult to find the discriminant boundary. (b) After kernel entropy maximization, the distances among samples are expanded in the feature space under constraints that hold only similar samples closely, making it easier to find the discriminant boundary.