HIGHER ORDER NEURAL NETWORKS
The aim of this session is the study of different types of "higher
order" artificial neural networks that may be included in heterogeneous
ANN architectures to improve the capabilities of classical neural networks.
New network architectures and their corresponding training algorithms,
based on the increase of synaptic processing, as well the processing power
of the transfer functions in the nodes will be considered. For example,
networks that incorporate the appropriate processing elements for the recognition
of patterns under different positions, orientations and scales, or networks
that allow the inclusion of the temporal dimension in their structure and/or
processing elements for time related problems. These new structures must
demonstrate their higher processing capabilities over traditional ANN architectures
with a reduction in the number of processing elements. They may be the
result of more or less biological inspiration, being the interest of the
session their applicability (with their training algorithms) to real problems.