Fig. 6
From: Discovering feature relevancy and dependency by kernel-guided probabilistic model-building evolution
Dependency heat maps in epistatic problems with 5 added noisy features. These matrices represent dependency graphs (gene–gene interactions) computed by Kiedra (notice that dependencies are non–directed since matrices are symmetric). Relevant variables are numbered starting from 0 (i.e. {0,1,2} in problem 3way, {0,1,2,3} in problem 4way and so on). Interaction intensity is represented by a warm–based gradient palette