what is the difference between feed forward and feed back methodology? Does the number of epochs in the neural network concept is the same defining the concept of back propagation methodology?

How to relate the concept when considering an image/ satellite image?

Answer 1

[Greg Heath]

Feed-forward nets are used when there are no physical or mathematical constraints on the order of data presentation. In particular, randomly scrambling the order of the data should not have a significant effect on the design and performance.

Feedback nets are used when there are physical or mathematical constraints on the order of data presentation. Typically, autocorrelation and/or cross-correlation functions have significant values at positive lags. In particular, randomly scrambling the order of the data should have a disastrous effect on the design and performance.

For this reason, the default 'dividerand' should be overwritten in the time-series functions timedelaynet, narnet and narxnet.

Number of epochs has no direct relationship to the comparison.

You have to be more specific regarding satellite imagery.