You do not have to read or understand this section unless you want to!
When it is being trained to recognize a font a Scan2CAD neural network is made up of three parts called “layers” – the Input Layer, the Hidden Layer and the Output Layer. See Advanced neural network information for a diagram.
The Output Layer is the set of characters that you are training the neural network to recognize. Each character you want to recognize is termed a “node”.
The number of nodes in the Output Layer
There are 256 characters numbered from 0 to 255 in the complete ASCII character range. Of these, there are 192 that are likely to appear on a drawing. The maximum number of nodes in the Output Layer is therefore 192.
By default the number of nodes in the Output Layer equals the number of characters represented in the training set. So, if you have entered examples of all the alphanumeric characters into the training set – A-Z, a-z and 0-9 – the number of nodes in the Output Layer will be 62 – one for each character.
You may want to train the neural network to recognize only a subset of the characters in the training set. For example, the training set may contain examples of all alphanumeric characters and you may only want to train the neural network to recognize upper case letters. In this case you need to restrict the number of nodes in the Output Layer to the letters A to Z only.
Restrict the number of nodes in the Output Layer
You can restrict the number of nodes in the Output Layer when you initialize a neural network.
The Initialize New Training Net dialog appears.
You can either set the value of each output node individually – e.g. node 1=A, node 2=B etc. or set the values of a range of nodes to a range of characters – e.g. nodes 1 to 26 = characters A to Z.
The first node in the range can be any node. For example if you wanted to train the neural network to recognize the characters a, b and c plus the characters A to Z, you could set the value of node 1 to a, the value of node 2 to b and the value of node 3 to c individually. You could then enter the range A to Z into node positions 4 to 29.
View the number of nodes in the Output Layer
To view the number of nodes in the Output Layer select Train Menu > Inspect Neural Net.
The Output Layer
Have questions on this topic? Talk to us