U.S. Geological Survey Open-File Report 02315
Abstracts for the Symposium on the Application of Neural Networks to the Earth Sciences
Donald A. Singer, Editor
INTRODUCTION Artificial neural networks are a group of mathematical methods that attempt to mimic some of the processes in the human mind. Although the foundations for these ideas were laid as early as 1943 (McCulloch and Pitts, 1943), it wasn't until 1986 (Rumelhart and McClelland, 1986; Masters, 1995) that applications to practical problems became possible. It is the acknowledged superiority of the human mind at recognizing patterns that the artificial neural networks are trying to imitate with their interconnected neurons. Interconnections used in the methods that have been developed allow robust learning. Capabilities of neural networks fall into three kinds of applications: (1) function fitting or prediction, (2) noise reduction or pattern
recognition, and (3) classification or placing into types. |
View or download this 24-page Open-File Report as an Adobe Portable Document Format (PDF) file (118 KB).
For questions about the content of this report, contact Don Singer.
Download Adobe Acrobat Reader for free.
| Download help | PDF help | Geopubs main page | Open-File Reports for 2002 |
| U.S. Department of the Interior | U.S. Geological Survey | Geologic Division | Western Mineral Resources |
This report is available only on the Web