Geoscience Reference
In-Depth Information
13.9 CONCLUSIONS AND OUTLOOK
CNNs provide much more than just a set of novel, useful or valuable data-driven mathematical tools.
Indeed, with respect to geographical data analysis and modelling tasks, they provide an appropriate
framework for re-engineering our well-established spatial analysis and environmental modelling
techniques to meet the new large-scale data processing needs associated with GIS and GC. The
application of CNN models to spatial data sets holds the potential for fundamental advances in
empirical understanding across a broad spectrum of geographical related fields. To realise these
advances, it is therefore important to adopt a principled rather than an
ad hoc
approach in which
spatial statistics and CNN modelling must work together. The most important challenges in the
coming years will be twofold:
•
To develop geographical application domain specific methodologies that are relevant to
neurocomputing
•
To gain deeper theoretical insights into the complex relationship that exists between
learning and generalisation - which is of critical importance for the success of real-world
applications
The mystique perceived by those outside the field can in part be attributed to the origins of CNN
systems in the study of natural neural systems, which, in association with the extended hype and
metaphorical jargon that is rife in this area of computer science, has acted to lessen the amount
of serious attention that is given to this new information processing paradigm. But - and this is
important to note - numerous aspects related to the subject of CNNs lend themselves to rigorous
mathematical analysis. This, in turn, provides a sound foundation on which to base an investigation
into the capabilities and limitations of different CNN tools and for examining their use in real-
world geographical applications. Casting such an analysis in the universal language of mathematics
would also be a worthwhile positive act that could help to dispel much unwarranted mystique and
avoid much potential misuse.
REFERENCES
Abdel-Moetty, S.M. 2010. Traveling salesman problem using neural network techniques. In
2010 The Seventh
International Conference on Informatics and Systems (INFOS)
, pp. 1-6, Cairo, Egypt.
Abrahart, R.J., F. Anctil, P. Coulibaly, C.W. Dawson, N.J. Mount, L.M. See, A.Y. Shamseldin, D.P. Solomatine,
E. Toth, and R.L. Wilby. 2012. Two decades of anarchy? Emerging themes and outstanding challenges
for neural network river forecasting.
Progress in Physical Geography
36, 4 (August 1): 480-513.
Abrahart, R.J., L.M. See, C.W. Dawson, A.Y. Shamseldin, and R.L. Wilby. 2010. Nearly two decades of neural
network hydrological modelling. In
Advances in Data-Based Approaches for Hydrologic Modeling and
Forecasting
, eds. B. Sivakumar and R. Berndtsson. Hackensack, NJ: World Scientific.
Abrahart, R.J., L.M. See, and P.E. Kneale. 1999. Using pruning algorithms and genetic algorithms to opti-
mise network architectures and forecasting inputs in a neural network rainfall-runoff model.
Journal of
Hydroinformatics
1, 2: 103-114. http://www.iwaponline.com/jh/001/jh0010103.htm.
Akgüngör, A.P. and E. Doğan. 2009. An artificial intelligent approach to traffic accident estimation: Model
development and application.
Transport
24, 2: 135-142.
Allouche, M.K. and B. Moulin. 2005. Amalgamation in cartographic generalization using Kohonen's feature
nets.
International Journal of Geographical Information Science
19, 8-9: 899-914.
Almeida, C.M., J.M. Gleriani, E.F. Castejon, and B.S. Soares-Filho. 2008. Using neural networks and cel-
lular automata for modelling intra‐urban land‐use dynamics.
International Journal of Geographical
Information Science
22, 9: 943-963.
Benjamin, C.O., S.-C. Chi, T. Gaber, and C.A. Riordan. 1995. Comparing BP and ART II neural network
classifiers for facility location.
Computers & Industrial Engineering
28, 1 (January): 43-50.
Bishop, C.M. 2007.
Natural Networks for Pattern Recognition
. Oxford, U.K.: Oxford University Press.
Brown, E., M. Aitkenhead, R. Wright, and I.H. Aalders. 2007. Mapping and classification of peatland on the
Isle of Lewis using Landsat ETM+.
Scottish Geographical Journal
123, 3: 173-192.
