Figure 3From: Noise-injected neural networks show promise for use on small-sample expression dataA visual comparison of different classifiers. The data model is a two-dimension nonlinear model. Sample size is 20, with 10 points from each class. The samples from class 0 and class 1 are denoted with round and cross markers, respectively. The shading denotes the decision region determined by the corresponding classifier, with darker shade for class 0 and lighter shade for class 1. The structure of the neural network is 2-4-2.Back to article page