Development Page--Not for Official Use

What Exactly is an Artificial Neural Network?

An Introduction to Artificial Neural Network Forecasting

An Artificial Neural Network (ANN), sometimes referred to as just a Neural Network, refers to a type of artificial intelligence that attempts to imitate the way a human brain processes information. A neural network functions through the creation of connections between processing elements which function as the equivalent of neurons. These connections are weighted such that a particular input stimulus will produce the desired output. Neural Networks are typically used when the relationship between the input and output is known to depend on several factors, but the interaction of those factors is not well known.

How Does an ANN Work?

There are two phases in neural information processing:

1. Learning or "training" phase and
2. Retrieval phase

In the training phase, a data set is used to determine the weight parameters that define the neural model. The weights are initially chosen at random and then the network is presented with a set of inputs that have known output. The output of the ANN is compared to this known output and the weights are adjusted to bring the Neural Network output closer to the actual output. This process is repeated until some criteria signify that the ANN has "learned" to recognize the pattern hidden within the data set (or there is no more data available to train the ANN). In other words, an ANN "learns" from previous examples (for example: as humans learn to recognize types of birds from examples of birds) and exhibits some capability for generalization beyond the training data.

Once trained, the ANN can then be used to process real data sets. The same types of input data are given to the ANN in the same order as during the training phase and the ANN outputs a value or values that closely mimics the learned relationship.

The ANN is particularly effective for predicting events when the networks have a large database of prior examples to draw on. There are no methods for training Neural Networks that can create information that is not contained in the training data.

Results of Neural Network Application to Water Level Forecasting

Links to More Information on Neural Networks

<< WhoMeasuresWaterLevelsAlongTheTexasCoast | ForecastsSiteMap | LinksToMoreInformationAboutArtificalNeuralNetworks >>