Another possible application could be to use a SECoS network to recognize the movements as described in the model scenario given in figure 8.4. A 'run' in this very simple scenario can be represented by a sequence of coordinates of those positions, which are occupied by the moving object. Without loosing generality it is for a demo run assumed that all runs are starting at point (This assumption can later on be relaxed, if the network has some flexible adaptation to different starting positions). One possible path is shown in the following example.
INP = 4. 4. 4. 5. 5. 5. 5. 5. 6. 5. 6. 4. 6. 3. 5. 3. 5. 3. 4. 3. 3. 3. 3. 4. 4. 4. 3. 4. 3. 5. 3. 6. 4. 6.
plot2d4(INP(:,1), INP(:,2), rect=[0,0,8,8])
These movements have to be associated with some labels. One simple solution would be, to map two consecutive points into one minimal movement label like
An output layer O could then have five output neurons and a desired output vector could have five elements either '0' or '1', like
No. | VECTOR-VALUE | MEANING |
1 | '0' or '1' | UP |
2 | '0' or '1' | LEFT |
3 | '0' or '1' | RIGHT |
4 | '0' or '1' | DOWN |
5 | '0' or '1' | NO MOVE |
A combined training input would then look like the following data set with the Dimensions
X Y U L R D N
INPM = 4. 4. 0. 0. 0. 0. 1. 4. 5. 1. 0. 0. 0. 0. 5. 5. 0. 0. 1. 0. 0. 5. 5. 0. 0. 0. 0. 1. 6. 5. 0. 0. 1. 0. 0. 6. 4. 0. 0. 0. 1. 0. 6. 3. 0. 0. 0. 1. 0. 5. 3. 0. 1. 0. 0. 0. 5. 3. 0. 0. 0. 0. 1. 4. 3. 0. 1. 0. 0. 0. 3. 3. 0. 1. 0. 0. 0. 3. 4. 1. 0. 0. 0. 0. 4. 4. 0. 0. 1. 0. 0. 3. 4. 0. 1. 0. 0. 0. 3. 5. 1. 0. 0. 0. 0. 3. 6. 1. 0. 0. 0. 0. 4. 6. 0. 0. 1. 0. 0.
Gerd Doeben-Henisch 2012-03-31