Simplified Information Maximization for Improving Generalization Performance in Multilayered Neural Networks

Figure 14

A variety of architectures by the information-theoretic method.