State of the Art and Development Trend - Computational Intelligence in Time Series Forecasting - page 324

10
State of the Art and Development Trend
10.1 Introduction
In the previous chapters we have presented the main issues of computational
intelligence that, in our opinion, are of outstanding interest to practising engineers
in the industry. The issues presented, to our knowledge, also make up the main part
of syllabus of postgraduate courses on fuzzy logic, neural networks, computational
intelligence, and soft computing in computer science and engineering.
In the last decade, some additional issues have become of growing interest,
such as support vector machines , wavelet neural networks and fractally
configured neural networks , which will be presented below. However, there are
some advanced issues like stochastic machines , neurodynamics , and
neurodynamic programming that have been left out of scope of the topic because
they are, in our belief, primarily of interest to informatics scientists and
mathematicians.
The new development trends in computational intelligence are multiple. In the
following, some development trends that, in our opinion, are the most promising
for applications in engineering will be presented.
In the area of neuro-technology, the progress and the expectations in the area of
bioinformatics and neuroinformatics are tremendous (Chen et al ., 2003). The
advances have prevalently been possible due to the availability of sophisticated
computer facilities for collection and management of tremendous amounts of
complex experimental data, required for analyzing of brain infrastructure (see the
Proceedings of the IEEE , special issues on bioinformatics, November 2002 and
December 2002). The foremost research goal here is to understand the synaptic
communication pathway of neurons and of supporting cellular elements. The
pivotal achievements thus far are the understandings of how the computational
processes in the living cells are performed by interaction of molecules and how the
stochastic biochemical networks are built. Although the research work for the time
being is predominantly carried out by biologists and bioneurologists, there is still
expectation that, at a certain point in development, it will attract the attention of