Time series modeling with pruned multi-layer perceptron and 2-stage damped least-squares method

TL;DR

This paper introduces a pruning method combined with a two-stage damped least-squares approach to improve multi-layer perceptron training for time series forecasting, reducing complexity and enhancing robustness.

Contribution

It proposes a novel pruning process integrated with a two-stage damped least-squares method for more efficient and robust MLP training in time series modeling.

Findings

Pruning improves forecasting accuracy.

Two-stage LMA enhances training robustness.

Method reduces training time and complexity.

Abstract

A Multi-Layer Perceptron (MLP) defines a family of artificial neural networks often used in TS modeling and forecasting. Because of its "black box" aspect, many researchers refuse to use it. Moreover, the optimization (often based on the exhaustive approach where "all" configurations are tested) and learning phases of this artificial intelligence tool (often based on the Levenberg-Marquardt algorithm; LMA) are weaknesses of this approach (exhaustively and local minima). These two tasks must be repeated depending on the knowledge of each new problem studied, making the process, long, laborious and not systematically robust. In this paper a pruning process is proposed. This method allows, during the training phase, to carry out an inputs selecting method activating (or not) inter-nodes connections in order to verify if forecasting is improved. We propose to use iteratively the popular damped least-squares method to activate inputs and neurons. A first pass is applied to 10% of the learning sample to determine weights significantly different from 0 and delete other. Then a classical batch process based on LMA is used with the new MLP. The validation is done using 25 measured meteorological TS and cross-comparing the prediction results of the classical LMA and the 2-stage LMA.