A neural network forecast demonstration
Advanced Search
Select up to three search categories and corresponding keywords using the fields to the right. Refer to the Help section for more detailed instructions.

Search our Collections & Repository

For very narrow results

When looking for a specific result

Best used for discovery & interchangable words

Recommended to be used in conjunction with other fields



Document Data
Clear All
Clear All

For additional assistance using the Custom Query please check out our Help Page


A neural network forecast demonstration

Filetype[PDF-2.96 MB]


  • Personal Author:
  • Corporate Authors:
  • NOAA Program & Office:
  • Description:
    Electronic computer implementations of artificial intelligence fall under two categories. In an "expert system" the programmer encodes a series of inviolable rules which the computer follows in making a decision. This approach requires a complete understanding of the problem to be solved. The computer is incapable of original thinking and unforeseen input patterns may cause the program to fail. Another approach to artificial intelligence, neural networks, can overcome these deficiencies when properly implemented. Neural networks (Jones and Hoskins, 1987; Stanley, 1988; Touretzky and Pomerleau, 1989) seek to emulate the structure and functioning of the human brain on the neuron level. The basic element of a neural network is the processing unit (Fig. 1). A processing unit usually receives input from several other processing units. These inputs are summed by a designated summation function (most often a simple arithmetic sum is employed). This sum is then put through a threshold function. Viewed simply, if the sum exceeds a certain predetermined value, then the threshold function allows a non-zero value to be sent out as output to other processing units further down the line. The links between the processing units are the heart of the neural network, for it is here that the learned "knowledge" of the network resides. Each link has a weighting value (usually between 0 and 1) that is uniquely its own. When the output of one processing unit is sent along a link to become the input of another processing unit, it is first multiplied by the weighting value of that particular link. During the network learning process, these weighting values are adjusted until the network has reached the required level of intelligence. Figure 2 demonstrates how actual values flush through a portion of a neural network. The input values of .5 and 1 are each multiplied by the weighting values associated with the inbound links to the processing unit. The results of these multiplications are summed at the processing unit and if they exceed the threshold value of that unit (which they do in this case) the sum is then sent out to the next layer of processing units.
  • Keywords:
  • Series:
  • Document Type:
  • Rights Information:
    Public domain
  • Compliance:
  • Main Document Checksum:
  • Download URL:
  • File Type:

Supporting Files

  • No Additional Files
More +

You May Also Like

Checkout today's featured content at repository.library.noaa.gov

Version 3.26.1