| Abstract: | Artificial neural networks (ANN) offer a supplement to traditional classification algorithms applied to archaeological data. Because of their flexible architecture, non-linearity and independence from the distribution of the underlying data, neural networks have unique advantages for such applications. In particular, ANN models are well suited for use with the sparse data sets common in archaeological work. Combining multivariate techniques with networks for data validation, pre-processing and classification exploits the merits of both and provides a comprehensive approach to the analysis and classification of archaeological data. |
