This book delivers theoretical and practical knowledge for developing algorithms that infer linear and non-linear multivariate models, providing a methodology for inductive learning of polynomial neural network models (PNN) from data. The text emphasizes an organized identification process by which to discover models that generalize and predict well. The investigations detailed here demonstrate that PNN models evolved by genetic programming and improved by backpropagation are successful when solving real-world tasks. Here is an essential reference for researchers and practitioners in the fields of evolutionary computation, artificial neural networks and Bayesian inference, as well for advanced-level students of genetic programming.

Offers a shift in focus from the standard linear models toward highly nonlinear models that can be inferred by contemporary learning approaches Presents alternative probabilistic search algorithms that discover the model architecture and neural network training techniques to find accurate polynomial weights Facilitates the discovery of polynomial models for time-series prediction Includes supplementary material: sn.pub/extras

Inhalt
Inductive Genetic Programming.- Tree-Like PNN Representations.- Fitness Functions and Landscapes.- Search Navigation.- Backpropagation Techniques.- Temporal Backpropagation.- Bayesian Inference Techniques.- Statistical Model Diagnostics.- Time Series Modelling.- Conclusions.