Regression Error Characteristic Optimisation of Non-Linear Models

In this chapter recent research in the area of multi-objective optimisation of regression models is presented and combined. Evolutionary multi-objective optimisation techniques are described for training a population of regression models to optimise the recently defined Regression Error Characteristic Curves (REC). A method which meaningfully compares across regressors and against benchmark models (i.e. ‘random walk’ and maximum a posteriori approaches) for varying error rates. Through bootstrapping training data, degrees of confident out-performance are also highlighted. This approach is then extending to encapsulate the complexity of the model as a third objective to minimise. Results are shown for a number of data sets, using multi-layer perceptron neural networks.