Problem Sequence

Problem Sequence

Top 

Problem Sequence

A Problem Sequence is made up of a base problem and all of its alias problems.  The Batch Processor indicates sequences by namable folder tabs in the main window.  The Batch Processor allows you to train multiple sequences of neural networks. Therefore, you may run more than one problem, with varying architectures and parameters, in order to find the best network for each problem.  In NeuroShell 2, this would be like training two completely different problems with many variations of each.

 

For example, suppose you wanted to predict the weather and electricity prices.  The collection of the base problem and alias problems for weather prediction would be contained in one problem sequence, while the collection of the base problem and alias problems for electricity prices would be contained in a second problem sequence.