|
Neural
networks are not a very usual item in the archaeologist toolbox.
They have the reputation of being powerful methods of data
generalisation, but computationally slow and mathematically
very complex. However, the availability of new software minimize
those problems, and allow the use of neural networks to many
other problems. In this paper, we apply supervised neural
networks (Backprop. learning algorithm) to the classical problem
of statistical hypothesis testing. Processing experimental
use wear in lithics we have found some contra intuitive results
using standard tests (chi-square, Cramer's V, Student's t,
ANOVA), which can be solved using the non-linear discriminant
power of Neural Networks. Specifically when archaeological
data do not fit parametric distributions, Supervised Learning
algorithms appear as an alternative approach. Our particular
case study is a set of digital images of experimental data
showing use wear as a result of work actions. We have used
experimental data in order to find similarities between use
wear identified in archaeological data and use wear whose
origin is well known from experimental results. Previous studies
shown that there is not an single discrimination rule to associate
cause (kynematics, worked material) and effect (use wear),
because of the stochastic nature of this phenomenon. Our paper
shows that Neural Networks can be an optimal method to classify
archaeological tools based on use wear
|
