Introduction to Multi-Layer Feed-Forward Neural Networks

Applications and Analysis of Multi-Layer Feed-Forward Neural Networks in Chemistry

Abstract

Basicdefinitions concerning the multi-layer feed-forward neural networks are given.The back-propagation training algo­rithm is explained. Partial derivatives ofthe objective function with respect to the weight and threshold coefficientsare de­rived. These derivatives are valuable for an adaptation process of theconsidered neural network. Training and generalisation of multi-layerfeed-forward neural networks are discussed. Improvements of the standardback-propagation algorithm are re­viewed. Example of the use of multi-layerfeed-forward neural networks for prediction of carbon-13 NMR chemical shifts ofalkanes is given. Further applications of neural networks in chemistry arereviewed. Advantages and disadvantages of multi­layer feed-forward neuralnetworks are discussed.