I am having in mind two alternatives: first, when an input vector produces an incorrect output, the weights are changed and it is checked that they work for the next vector.
Or, if the input vector produces an incorrect output, iteratively change the weights until a correct output of that vector is found and then move on to the next one, to see if the new weights, suitable for the previous vector, are good for the new one. .
What do you think? Is the last one a correct procedure? Any efficiency, or reliability concern for any of them?
- Badges
- Science topic
- Similar topics
- Vectorization
More Rey Segundo Guerrero-Proenza's questions See All
Similar questions and discussions