We present methods specially designed to be effective with moderately non-linear regression relationships. The model fitted is of the Projection Pursuit Regression (PPR) type with a smooth, non-parametric link function connecting the mean response to a linear combination of the regressors. New algorithms, close to ordinary linear regression, are developed. Considerable numerical evidence is giv...

Product models of low dimensional experts are a powerful way to avoid the curse of dimensionality. We present the "under­ complete product of experts" (UPoE), where each expert models a one dimensional pro­ jection of the data. The UPoE may be inter­ preted as a parametric probabilistic model for projection pursuit. Its ML learning rules are identical to the approximate learning rules proposed ...

We are developing a forecaster for daily extremes of demand for electric power encountered in the service area of a large midwestern utility and using this application as a testbed for approaches to input dimension reduction and decomposition of network training. Projection pursuit regression representations and the ability of algorithms like SIR to quickly find reasonable weighting vectors ena...

A new multiresolution algorithm for image compression based on projection pursuit neural networks is presented. High quality low bit-rate image compression is achieved first by segmenting an image into regions of different sizes based on perceptual variation in each region and then constructing a distinct code for each block by using the orthogonal projection pursuit neural networks. This algor...

Combining the advantages of genetic algorithm (GA) and projection pursuit regression (PPR), this article firstly uses improved Hermit polynomial as the ridge function of projection pursuit regression model. And then adopted the real-coded genetic algorithm to optimize the projection direction, a forecasting model for peak flow of short flood forecasting is presented. Applied the presented model...

We extend a reinforcement learning algorithm which has previously been shown to cluster data. We have previously applied the method to unsupervised projection methods, principal component analysis, exploratory projection pursuit and canonical correlation analysis. We now show how the same methods can be used in feature spaces to perform kernel principal component analysis and kernel canonical c...

This paper introduces the R package FKSUM, which offers fast and exact evaluation of univariate kernel smoothers. The main computations are implemented in C++, wrapped simple, intuitive versatile functions. based on recursive expressions involving order statistics, allows for smoothers at all sample points log-linear time. In addition to general purpose smoothing functions, built readyto-use im...